My response to all the FS5 artefact issue complainers.

FS5-artefacts-cap-1024x660 My response to all the FS5 artefact issue complainers.I’m getting a bit fed up with this. So lets just clear a few things up.

PLEASE ALSO SEE THIS RELATED POST WITH SAMPLE FOOTAGE AND FULL RESOLUTION FRAME GRABS. https://www.xdcam-user.com/?p=4583

SEE SONY’S OFFICIAL STATEMENT ON THIS ISSUE.

Native ISO: The Sony PXW-FS5’s native ISO for the standard gammas and cinegammas is 1000 ISO. It is NOT 3200 ISO. If you shoot with standard or cinegammas at 3200 ISO then you are adding +10dB to +12dB of gain.

+10dB gain = noisy picture. If you add +10db of gain to most cameras the picture will get noisy and grainy, the FS5 is no exception to this. Adding +10db gain also means you will have a +10db noise increase (roughly 300% more noise, a significant increase) or have an increase in noise plus a significant increase in other artefacts as the cameras built in noise reduction has to work much harder. The FS5 does incorporate noise reduction processes and these do introduce a degree of smear when you add more than +6db of gain. At +12db or higher the smear and other artefacts (including a tendency towards banding) becomes quite noticeable, this is a typical artefact of this type of noise reduction and one of the reason why on more advanced cameras like the PMW-F5/F55 you are able to turn it off.

Ultra Compressed: In UHD the XAVC-L codec has a bit rate of 100Mb/s. Uncompressed UHD requires over 3.5Gb/s for decent quality, so you’re looking at a camera with a compression ratio of around 35:1. Is it really all that surprising that there are some compression artefacts? Consider that 35Mb/s is considered the minimum for broadcast quality H264 based HD work and that’s a compression ratio of only 25:1. So we are some way below the normal minimums for broadcast.

8 bit in UHD. UHD XAVC-L is an 8 bit codec. 8 bit codecs don’t deal with noise terribly well as the limited number of shades/steps mean that noise cannot be reproduced with small steps and as a result any noise or grain will often appear quite coarse. So is it any surprise that in UHD the camera exhibits a lot more artefacts than in HD where the codec is 10 bit and much less compressed?

S-log at high ISO’s. Oh come on people, please understand that S-log only does one thing well and that’s maximise dynamic range. If you can’t get a correct exposure or better still a 1 to 2 stop over exposed exposure at the native ISO you really, really shouldn’t be using log. It will be noisy, it will be grainy, it won’t look good and if you record it with XAVC-L in UHD it will look like rubbish as the excessive noise will stress the codec, introduce more artefacts and your exposure mid range will only represented by approx 17 code values or shades. you will have very little data to play with in post and noise and grain will look awful.

External Recorder. Adding an external recorder can really improve the UHD image quality. Again, this shouldn’t come as a surprise. If you use an external ProRes recorder running at 880Mb/s (ProRes HQ) compared to the 100 Mb/s of XAVC-L is it really any surprise that it’s possible to record a better quality image? The FS5 has been designed as a grab and go camera recording on to cheap media. Sure adding an external recorder can increase the image quality, but your media costs go through the roof and most of the grab and go benefits are lost.

So when you see a test of XAVC-L with a standard gamma at 3200 ISO being recorded using an 8 bit,  35:1 codec don’t be surprised to see a noisy, grainy image with compression artefacts from all the extra work the codec is having to do to deal with the noise that comes from adding +10db gain. The same for low light log footage at 6400 ISO. This is not the best way to use a camera like this. It’s not clever and anyone can make even a straight forward scene look like rubbish by shooting this way. I don’t know what people are trying to say or achieve when they post a bunch of high gain clips with headline titles like “The FS5 codec is broken” or “FS5 un-usable in UHD” without even considering what it is that they are actually looking at. AT 3200 ISO with standard gammas the camera noise reduction circuits are working overtime to try and clean up the image. This results in the introduction of other artefacts such as edge tearing or smear. Stick that into a super compressed codec and it is not going to be a perfect image. 50Mb/s HD at 60fps is also very highly compressed and will also exhibit artefacts.

As I have written in other articles, the use of ISO with video cameras appears to be miss-leading many people into thinking that a camera will produce a noise free image at all kinds of ridiculous sensitivities as they often focus on the wrong ISO rating or simply believe that it’s possible to have very high sensitivities without noise. Sadly this is not the case. I don’t think people would be surprised to see noise and grain at +10dB gain, but sadly dB’s of gain isn’t hip, cool or make you sound like a film cameraman. No, ISO is much cooler sounding, but is confusing the c**p out of people that don’t really understand how it works in relation to a video camera. Also the use of log for low light is just nuts, it’s entirely the wrong type of gamma to use, especially with a sub optimum codec.

The FS5 is not simply a shrunk down FS7. It is a very different camera. You should not be expecting FS7 performance in UHD, the UHD version of XAVC-L codec is very different to the XAVC-I available in the FS7. The image processing is also different (do remember the FS5 consumes 1/2 the power of an FS7).

300x250_xdcam_150dpi My response to all the FS5 artefact issue complainers.

It is possible that over future firmware updates Sony may be able to fine tune the codec and noise reduction circuits to work a little better. It’s also possible that we may see improvements in the decoders used to decode the codec (remember the decoder is just as important as the encoder), so possibly things may get improved a little. But don’t expect miracles. Squeezing UHD into a highly compressed 8 bit codec and recording it 100% reliably on off the shelf SD cards will always be a challenge. Even at the correct native ISO’s, at 0dB gain, there will be some artefacts. But start adding in gain and yes, you will start to see more noise and more artefacts.

The FS5 is a great little camera, I really enjoy shooting with mine and I think the results I am getting are great. I know that I can get a technically better 4K/UHD image from my FS7 or F5, but sometimes it’s not just about getting the best technical quality. The FS5 allows me to shoot more freely, it’s a breeze to carry around or travel with, I can throw it on a gimbal, on the end of a microphone boom pole, chuck it and a bunch of lenses in a small back-pack, it’s fun to use! As a result I’m getting shots that I just can’t get with the FS7 or F5.

Just how terrible is the FS5’s image quality? Take a look at my Falcon video and judge for yourself.

Northern Lights 2016: I still have 2 places left on my February expedition to Norway to the Finnmarksvidda, land of the Sami people and the Northern Lights. Full details here.

460x150_xdcam_150dpi My response to all the FS5 artefact issue complainers.

133 thoughts on “My response to all the FS5 artefact issue complainers.”

  1. I’m so happy you addressed this. Twitter has been crawling with complaints and misunderstanding the last few weeks

  2. I believe the edge tearing artefact reported by some is due to the temporal noise reduction that the camera uses. As you pan or tilt the camera the areas around edges will change from frame to frame while large more solid areas of the picture change little. Because of the differences from frame to frame you won’t get any noise reduction in these areas so edges can become coarse and “fuzzy” or jittery as there is on the one hand a marked increase in noise on the edges and at the same time a ghost image from the adjacent frames. At 0db this is barely visible and certainly not an issue with real world shots but ramp up the gain and this increases rapidly. These are fairly typical side effects of temporal NR.

    1. Glad you’re acknowledging the edge tearing issue. I agree that it’s most likely temporal noise reduction. But it is poorly tuned on the FS5. Never seen it like that on any other camera. Thats why I’ve been talking a lot about it, because I hope Sony can tweak it. You’re attitude of “just deal with it” is unfortunate in this post.

    2. Glad you’re acknowledging the edge tearing issue. I agree that’s it’s most likely the camera’s temporal noise reduction. But it is poorly tuned on the FS5. Never seen it like that on any other camera. That’s why I’ve been talking a lot about it, because I hope Sony can tweak it. Your attitude of “just deal with it” is unfortunate for those that want Sony to make adjustments to the NR to reduce the edge tearing problem. +9db isn’t an unrealistic or “not real world” setting. Plenty of documentary shooters shoot indoors with those settings.

      1. FFS. Adding that much gain adds 3x the noise and or NR artefacts. That’s a fact of electronic life, you can’t change that, that’s the physics. When cameras used to only have gain you didn’t get stupid remarks or silly tests with people complaining that “Oh dear, my cameras noisy when I add a ton of gain”. We are sadly living in an age of lazy shooters that expect their cameras to be able to see in the dark that either can’t be bothered to light or don’t know how to light. I would suggest you need to look at a lot more cameras with 10 to 12 dB of gain added and look at how the NR behaves on those cameras. It’s operation is normally quite clear to see as either image smear or image softening. It should not be a surprise to anyone that there is a noticeable increase in artefacts when you raise the gain levels.

        1. Quit the condescending attitude Alister. We’ve all seen the other cameras and they don’t do it. The A7R II has a 15MP pixel count across the Super 35mm crop area of the 42MP sensor, that’s a more dense pixel pitch than the 11MP Super 35mm sensor of the FS5. You can shoot at 10dB+ gain over the base sensitivity of the image profiles on that camera, you can even go happily up to ISO 12,800 in S-LOG 2 on that camera, WITHOUT the same kind of fuzzy edge image processing flaw everyone and their cat is seeing on the FS5! It’s not about a noisy image. It’s about a broken image!! You’ve seen the proof haven’t you? Surely you have watched the evidence on Vimeo and downloaded the MXF files which show the issue very clearly. Take your fingers out of your ears and start listening.

          1. When are you going to understand that when you mix different sensors with different processors you will get different results. You just go ahead and expect miracles. You just keep on shooting log in low light as you clearly have no clue about the way log gamma works. Just chuck all that extra gain in there and ignore the inevitable consequences on the image.

          2. I have to defend Alister here. I am not specifically referring to you alone here Andrew, but here seems like a good place to post this.

            Alister appears to be bang on with all his explanations for faults based on usage. Sure there seem to be some points not completely clarified, and time will tell if Sony are able to/chose to improve them.

            But, the bottom line is that a lot of people seem to be complaining based on comparisons with other cameras, namely the A7 models.

            Just because two cameras are made by the same manufacturer doesn’t mean that all components/signal processing chain will work in the same way. Doesn’t even mean they should be priced to match.

            Sure, as a consumer that would be great, but for whatever reason that is not how a huge company like Sony has chosen to operate.

            I’m pretty certain the A7 camera, the FS cameras and the Cinealta cameras are designed and made by totally different divisions of Sony. Sure, marketing group them together, but they are from different sources. Heck, Sony’s sensor division themselves are totally separate. Otherwise we’d see that 100mpx Phase One sensor in a Sony camera, and the A7S sensor in Cinealta cameras.

            Sure, I’d love my PMW-F5 to have a sensor as sensitive and clean at high ISOs as the significantly cheaper A7S. But that’s not going to happen. In the same way, the FS5 shouldn’t be compared, and really it is very different from it’s bigger brother too. Regardless of relative pricing.

            As a long time Sony user and very close follower of their products, developments, quirks there is one thing I know and that will unlikely ever change: a lot of stuff related to inter product comparison within the brand doesn’t make sense. It’s no different for other brands like Canon (how many DSLRs are actually needed?!).

            Maybe not every one and every example is to blame, but usage really does make A LOT of difference to how a camera works. That is something I have learnt (and try to teach) more and more since transitioning to a world of pure video camera to more “cinema style cameras” just over two years ago.

            Stop comparing and think of each product (or small groups of products) as separate entities. I wish it wasn’t this way… but it is.

          3. For those that have used “Neat Video” which is a fantastic noise reduction plugin…

            Have you ever sampled a bad noise print and tried to aggressively apply that model to your video? You get problems that look a little like what the FS5 is doing today.

            Allister, I think your assessment about this being a temporal noise reduction problem could be spot on. Is it possible that it’s simply an aggressive use of a “bad”, “incorrect” or “improper” noise model? (Sony, like ALL other cameras companies, have a track record of making big mistakes in firmware…ie, A7s-II black hole sun just to name a recent one)

            Does Sony build one ASICs noise reduction chip that it can use for multiple sensor types? Could this chip contain maybe 5 or more noise models based on popular Sony sensors? (which they can also sell to Sony sensor buyers) Could the FS5’s firmware be accidentally calling the wrong noise model? (i.e. It’s calling the FF sensor noise algorithm instead of the proper S35 algorithm)

            I don’t know, I’m only throwing out guesses.

            We all know what good noise reduction looks like. My guess is that we are now seeing what bad noise reduction looks like. OK, maybe not “bad” noise reduction but “wrong” noise reduction, instead?

            CT

        2. Alister – your attitude about this is very sad. I for one used to look up to you but am now seeing petulance that is very unbecoming. The NR is poorly implemented on the FS5. You keep redirecting the discussion into what gain is acceptable to use. Increased noise or softening is fine because it lets the user deal with it the way they want. But the irregular artifacts that the FS5 has at +9db takes shot that could have been cleaned up in post and makes them unrecoverable.

          If an editor applied a NR filter to a shot and those artifacts showed up, then the editor used the NR filter poorly and would be asked to tweak the temporal sensitivity parameter. There’s nothing else to it. No need to call people lazy or stupid because they see it as an issue. This is all a bizarre episode from you.

          1. To be fair Max, when this whole thing originally started (not on this blog post, but elsewhere) a lot of it was down to things like high ISO clips, noise floor etc etc. The issues regarding noise reduction came later.

            Also, isn’t it the case that any noise reduction issues are primarily seen in instances with a high noise floor/high levels of gain?

    3. Hi Alistet. I own the camera and I can assure you that the very obvious edge tearing only happens in slow tilts but not in pans. It is clearly directional. Before becoming a video artist I worked in software development and I know enough about the algorithms involved in noise reduction that this is a very unusual behavior to say the least. I’m pretty sure this is a fault in the implementation. Not sure if it affects all cameras but on mine it is harsh and nothing if seen on any other high-end or consumer camera I ever worked with. If you can’t reproduce it I’m happy to provide you with a detailed analysis. I Iove the camera for many reasons but as things stand at the moment I can’t use it the way intended.

      1. It is very, very obviously a temporal noise reduction artefact. You don’t need to be a rocket scientist to see what is happening. As the camera is tilted the NR is reduced along any hard edges to prevent excessive smear, the side effect of this is a significant increase in noise in these areas. I’m sure it could be tuned differently, perhaps a softer picture or more smear would be preferable? But however it’s tuned the fact of the matter is that it’s NR. Use the camera without adding gain and the problem is invisible in normal situations. Thats how the camera was intended to be used, not with +10 to +12dB of added gain. If that was how it was intended to be used then the camera would have been set up with native ISO’s of 3200, not 800/1000.

        1. Hi Alister (sorry for misspelling your name before, I was typing on a phone). What I’m experiencing are noise reduction artifacts, of course. My point is that these artifacts only show in vertical movements (eg. a tilt) but not in horizontal movements (pans). With horizontal movements the noise reduction seams to work flawlessly. So how come there is this discrepancy? A (likely motion vector based) NR algorithm “looks” in all directions without “discriminating” one direction over an other. (And the camera is not line-skipping by any chance?!)

          Additionally these artifacts look like nothing I’ve ever seen on a camera before, not even pushed to extrem ISO (or gain) settings. What makes it so problematic for me is the fact, that opposed to noise, grain, smearing and all the other effects of high ISO (or gain) settings, these blocky artifacts are so obviously digital image defects. We are used to see noisy images (blown up 16mm for a start) and we are used to see smeared video footage (HDV night time news coverage) but when we see blocky pixel groups jumping around we (and the audience of our work) can only think of a technical defect.

          So, I’m pretty sure it is a fault in the implementation of the algorithm and I very much hope there is something Sony can do about with an update. (Otherwise I’ve just lost quite a bit of money and learned a lesson.)

          Thank you for dealing with this btw. It seams all a bit messy.

          1. Perhaps Sony feel that most as most camera movement occurs in the horizontal axis with typical motion such as pans, left-right motion of objects through the shot etc that by using more vertical based NR they can provide a lower noise image without having excessive smear or blur on the more common horizontal motion. I really don’t know, but it may be by design and I would not assume it to be a fault just because it’s most noticeable in one axis. Temporal NR is rarely vector based, it’s normally just time based. Take a single pixel and over 2 or more frames take an average value and blend that with the original value to dramatically reduce the noise. But it adds a lot of smear.

            But the bottom line is you only get this “issue” when you use excessive amounts of gain. Perhaps Sony will “fix” the problem, perhaps we will get equal amounts of the artefact in both horizontal and vertical axis, or perhaps they will reduce the overall NR and you’ll just have to put up with noisy pictures and more compression artefacts?

    4. Hi Alister, really appreciate your responses in helping users with their problems 🙂 Have no idea how you shot The Falcon with so clean images! Mine at 0db gain creates edge tearing artifacts even in 10-bit HD at native ISO in outdoors daylight conditions 🙁 Can’t understand what’s wrong….

      1. I don’t understand why other people seem to be struggling so much with this camera. Are you sure you are really at 0dB because at 0dB my camera produces a wonderful image and that’s what most other owners are finding. The tearing only occurs when extra gain is added.

  3. Alister this is all well and good. But why then do the 422, 50mbit 10 bit color HD images show the same horrible banding and artifacts in a brightly lit sky? This is not an issue of high ISO, or low light, or color depth, or not having enough data. The HD artifacts are just as bad as the 4K ones in many scenes. This is not about telling everyone they don’t know what they are talking about. Or that they shouldn’t be using the camera as designed it advertised. I find it frankly absolutely preposterous that you are now telling people essentially “don’t shoot at ISO in Slog” when that’s the only choice. Ok then “don’t shoot 4k because it’s 8 bit low bitrate” when the artifacts are the same in the much better speced HD. Or “light your shadows” when the sky brightly lit by the sun, a fairly big light source last time I checked, does nothing to hide the artifacts. These faults in the camera are inconvertible and factual and self evident in every single recording showing large areas of color or shadows that i have seen. Blaming the user for “not doing it right” is disingenuous at best.

    Your study of the math and science of how slog and all of this works is indeed sound, if not entirely artistically relevant to artists who want things to “just work”. However it’s also not entirely applicable in this case as the cameras performance is contrary to you own characterizations about how it should perform given the bitrate and color to flourish in. Sony is selling a camera with clear flaws that do not appear in any of its other similar large sensor cameras. Educating people what they are truly getting into, versus blaming everyone who has a legitimate complaint, is doing service to artists where $5600 is a lot of money.

    Unlike you, or most of the rest of us, we are not paid by or sponsored by Sony. There’s zero conflict of interest on my sure of the fence. Deflecting or watering down a clear flaw in the image in this camera by telling seasoned camera people and those starting out alike they don’t understand how a camera works is not productive. It may lead some down the wrong path when their client then complains about the images afterwards.

    I have said many times I actuallylike the camera a lot despite these flaws. The body design in particular is a dream. But the images are flawed as compared to all of Sony’a other large sensor cameras and need work to fix, if fixable at all (eg smearing). People must not go into buying this camera thinking they can fix it by “proper” shooting or they may be sorely disappointed. I would not hire an FS5 on any shoot without demanding a recorder be used with it in any but the most lenient of circumstances. I would not want my shoots polluted with that artifact filled, posterized garbage. It’s only the fS5 that performs like this. Not the A7RII or A7S or even the RX100. Only the FS5 is making such a mess in the images probably due to as you state poor or over aggressive noise reduction. Let’s please own up to the facts and stop trying to make excuses for what is self evident in the shots from many people, least of all myself. You of all people, who shoots mostly skies, should know better.

    People can make decisions on whether to purchase based on the footage itself. Pointed at a brightly lit sky. In glorious 50 Mbit 422 10 bit HD. Full of banding and blocky messy artifacts.

  4. I was as excited as anyone else for the FS5. It seemed almost as perfect a camera for me as could be, but these image issues are a dealbreaker. Clearly the FS5 is aimed squarely at the C100II, yet it seems to not even match the Canon in high-ISO noise or lack of silly artifacts… let alone beat it! That’s a hard sell, especially now that the C100II has had a price drop. Sony better clean up the FS5 mess quickly, with a free firmware update, or it’ll forever be out of consideration for lots of people.

  5. Alastair, To answer your headline question, no I do not. But I expect an awful lot better image quality from a professional video camera costing $5600. In my humble and unscientific opinion, my lowly RX10 II and NX1 shoot better 4K.

  6. Thank you very much, Alister, I have been trying to point out that ISO = Gain multiple times but it might be hard to grasp. I also stated in my test that it is not happening at all at the individual base ISO / Gain of the available Gamma curves.
    I was also mentioning that the XAVC-L codec is not the problem and that the edge tearing we are seeing is noise “disguised” as macro-blocking (I had the suspicion that it is some kind of internal noise reduction).
    Now comes my question to you: How is it possible to turn off the temporal noise reduction? I have no problem with noise but I would rather prefer to have it removed in post. As a side note I just want to mention that my F5 is behaving way better at similar settings (I know that the codec is a different one but still, independent of noise reduction settings etc.).

    Despite all that I am very happy about the FS5, it’s exactly what I wanted to have… something with comparable performance (to the F5) but way easier to set up, transport and operate.

    1. You can’t turn off the NR. I suspect that if you could the noise would swamp the codec. The F5 which uses the same sensor has the ability to turn off some of the NR. With the NR off at 1000 ISO with standard gammas the F5 is twice as noisy as the FS5. You can just about get away with 100Mb/s in HD or 250Mb/s+ in UHD with XAVC-I, but don’t think it will work with XAVC-L especially with 8 bit UHD. Long GoP codecs really hate noise.

      1. Alright, then please tell your Sony contacts to make the NR switchable so those who want to record it or maybe via the HDMI / SDI out can deal with it anyway they want.

  7. Agree with everything Paul Antico says. +100. People are not ‘complaining’ they are ‘reporting’ and giving feedback. People who have bought this camera for pro use are very disappointed.

    HD 50mbps 422 should not have gradation, banding and artefacts like we are seeing. 4K 8bit 420 YES.

    And if it is a Noise Reduction issue I’m sure Sony can implement a firmware to turn NR off no?

  8. Yes, the over-aggressive temporal noise reduction on the FS5 is a problem. I wish you wouldn’t try to brush it off or downplay it. Sony bills this camera as “grab and shoot” and as a great camera for documentary. For documentary and wedding videographers, shooting at +9db is an everyday occurrence. All I want to do is gain exposure for the the edge tearing artifacts in the hopes that Sony can tweak the noise reduction parameters to reduce the artifacts.

    1. Maybe for you shooting at +9dB is a regular occurrence, but that does not change the fact that adding +10dB or more of gain will always increase noise and/or other artefacts 3 fold.

  9. I don’t own an FS5 yet. (Getting mine at the end of January….so I can’t say much because I can’t test it myself yet. However, many complaints center arround an idea that this particular and specific horizontal edge tearing artifact does not appear to exist on any other Sony camera. It does not itself like this on the A7S/R-II, the Rx10-II, X70 or the FS7. It’s also something that seems to be “pre codec” and also exists on the HDMI feed too.

    Some have gone as far as to speculate that the FS5 implements the A7S-II noise reduction ASIC chips that were not designed for the FS5’S S35 sensor. They have claimed that the FS5 was rushed to market with the wrong noise reduction algorythm modeling being implemented.

    Again, it seems that we can’t blame the FS5’s codec on this one as the problem happens way before it’s presented to the codec.

    I don’t know what to think until I test it myself. I believe that testing the FS5 against the RX10-II and the A7S-II is a fair comparison. The FS5 should perform in the middle of the two. THE FS7 has the same sensor and it’s 8bit HDMI output “should” look close to the HDMI FS5 output. However, if the lowly RX10-II can out resolve the FS5 with slow movement on a resolution chart?….that’s going to be a big problem for me.

    The noise and splotchy edging issue on the FS5 today is very specific to horizontal lines and does not look like the typical noise reduction side affects that I have seen a million times before.

    I don’t claim to know anything right now. I’m only going off other people’s samples today.

    I’ll know for sure in 3 weeks!

    CT

    1. The FS5 IS NOT a shrunk down FS7. It is a VERY DIFFERENT camera with totally different image processing. The FS5 is much closer to the FS700 than to an FS7. To expect the FS5 to perform the same as an FS7 is a mistake and I’m sure this expectation is at the heart of many of these complaints. People that have purchased the much cheaper FS5 but expected FS7 performance.

      1. Thats your answer? You have backed yourself into a corner without obviously testing.

        Nobody have mentioned wanting the same performance as a FS7.

        Will you acknowledge there is an issue even in HD at least?

        1. I own an FS5 and have done for almost 2 months and have shot extensively with it. My FS5 short “The Falcon” is on Youtube if you wish to see the performance I get from it, take a look for yourself.

          Yes if you slap on a ton of gain the image quality drops, but that is to be expected.

        1. Then buy the FS7!

          But don’t forget the FS7 will need a bigger tripod, more expensive media, more batteries. The real difference in cost of ownership is much greater than £600.

  10. Let me preface this question with a statement- I am making the transition from still photography into videography and I don’t claim to be an expert. Most of what I know about S-Log and gamma curves in general comes from you.
    Can you explain why the base ISO for S-Log is 3200 on the FS5? Isn’t that kind of crazy considering that S-Log is meant to be used for scenes with a wide dynamic range such as bright, sunny days?
    This seems especially strange considering that the FS7, which I’m told uses the same sensor, has a base ISO of 2000 in S-Log. I’ve heard people say, “Well, the FS5 uses different processing” but that seems like a cop out and doesn’t really explain the technical reason for bumping up the gain so much. People also say that the A7S also has a base ISO of 3200 in S-Log, but it also has much larger pixels so that level of gain doesn’t seem to be so extreme.

    1. The FS5 has no ND filter in front on the senor, where the ND filter system is raised clear of the sensor and ND ON where the variable ND is in front of the sensor. Once the ND is in place you then have 3 preset ND settings. The FS7 has a clear ND in front of the sensor that is why the base ISO differ.

      1. So what exactly is a “clear ND filter”? It’s either a clear glass with no appreciable light attenuation or it’s an ND filter. In the clear position there is no appreciable light attenuation on either the FS7 or the FS5. The ISO difference is due to processing differences. Even if you add the same amount of ND on both cameras the ISO of the FS5 is still higher than the ISO of the FS7.

    2. Dynamic range is not just about highlights or bright sunny days. It’s about RANGE from the darkest deepest shadows to the brightest highlights. One of the reasons the base ISO is higher in log mode compared to standard is because of the greatly extended low key or shadow range, the camera is able to see much deeper into the shadows, that is why it is more sensitive in log mode. But that normally comes at the expense of more noise and grain as you are seeing further down into the sensors noise floor, dark parts that would not be seen with standard gammas. However using noise reduction can help mitigate this and different processors can apply different amounts of NR. The FS5 is a very different camera to the FS7. While the sensors are the same, the image processing is very different. Perhaps the image processor in the FS5 handles shadows better than highlights so to get the 14 stop dynamic range it has to operate at a higher ISO.

  11. So Alister,

    Are you putting your reputation on the line by endorsing a faulty Sony FS5 without actually investigating?

    This is HUGE this could pur Sony out of the picture.

    1. Do you think I haven’t looked very closely at the images from my own FS5?

      FACT: 3200 ISO with a standard gamma or cinegamma is the addition of between 10dB and 12dB gain. That’s a significant amount of additional gain.

      FACT: Adding that much gain with any camera will treble or quadruple the amount of noise or artefacts.

      FACT: S-Log is an appalling choice for low light as you are either wasting the majority of your recording data or forced to add a lot of gain to get the image into the “fat” part of the curve.

      FACT: 8 Bit UHD at 35:1 is going to have visible artefacts.

      FACT: 50Mb/s HD at 60fps is going to have visible artefacts.

      FACT: The FS5 is not an A7RII or any other camera, so it will perform differently.

      1. “FACT: 3200 ISO with a standard gamma or cinegamma is the addition of between 10dB and 12dB gain. That’s a significant amount of additional gain.”

        This is certainly so. However, it seems to me that in other Sony camera’s, they have managed +10-12db of gain much better and without these NR flickering edge problems. Have you seen the RX10-II? That little tiny 1 inch-type sensor at +10db doesn’t have edges that pulse or fluctuate like these FS5 samples do. I’m sure the signal to noise ratio of the FS5’s S35 sensor must be far greater than Sony’s 1inch-type model. Sony has excellent NR algorithms and techniques in it’s technology chest that can handle +12db better then this FS5 does today.

        In my opinion without having my own FS5 yet to be sure….I think something went wrong in the FS5 design lab. This camera is not processing NR properly and it’s not up to Sony’s normal standards. There is something (some oversight) in the firmware that Sony needs to revisit.

        We all know gain hurts every camera’s image. It brings up the noise floor of the sensor. The NR circuit tries identify what is grit and garbage and tries shove the floor back down without hurting the parts of the waveform that our eyes want to keep. That’s all normal and par for the course as they say.

        However, to me, this artifact is very odd looking for +10-12db. Something is wrong today and when Sony fixes it, we’ll all say:

        “OK, now THIS is the way that PROPERLY soft edged and “water-colored” noise reduction should look on the FS5 at +12db…”

        Suggestion: How about a NR strength option? Variable options 0 thru 5. I use Neat Video ALLOT and on some projects, I’d rather capture more detail (or in HDMI with ProRes) in camera and work carefully with Neat Video in post.

        1. Yes, the ability to alter the NR levels would be nice, but my suspicion is that if you reduce the NR below where it is now you will start to swamp the codec. It is a very fine balancing act to perform when dealing with high compression rates. More in camera noise will result in not just more noise but also softer images as the H264/XAVC encoder also must balance sharpness with artefacts.

          I would suggest you wait until you have tested an FS5 for yourself before passing judgment on the effectiveness of the NR. My experience is that it does a very good job in most typical situations. Sure you can find very specific ways of finding flaws in it, but I have yet to find any form of NR that doesn’t have flaws of some sort.

  12. Alister, don’t want to add add more fuel to the fire but I think you might want know that EOSHD has posted a direct rebuttal to your comments about this issue:

    http://www.eoshd.com/2016/01/response-alister-chapmans-response-sony-fs5-complainers/

    In the past 10 days, this FS5 problem has developed allot of internet “buzz” or concern. Several industry web sites and groups are discovering the problem and very few people that see it think it’s “normal” or OK.

    I think this is a very urgent issue that Sony must investigate and tackle right away. I don’t believe that this will just “fade away” on it’s own over time and we will all just forget about it. In fact, I’m certain within the next month, we will see dozens of angry people posting more and more videos with torn and ragged edges.

    I certainly plan on buying the FS5 and getting it on my resolution carts on day 1 with the A7S-II, RX10-II and X70.

    I was concerned that Sony engineers in Tokyo might not be aware of this problem yet but now that you have posted a response on this topic, I’m certainly sure they are 100% fully aware now.

    I’m a HUGE Sony fan and I have owned 16 Sony cameras in 20 years. I’m fully confident that they will fix this issue soon.

    CT

    1. Yes I’vs seen this and it’s full of incorrect or sensationalist statements that really do show how little Andrew understands video cameras.

      1. Sorry, Alister, but Andrew Reid does know what he’s talking about and certainly represents voices like mine. On the strength of opinion that’s out there on the FS5, I am going to be changing my recommendation as to the next camera purchase my company is about to make. The FS5 sounded ideal for our purposes – but in the end picture quality is what you’re ultimately paying for. And if I now have to add an extra step when setting up the camera in the hope that I avoid exposing its problems, then I’m not interested. I’ll do my bit, creatively, Sony can do its bit on the technical. Just get it right – first time.

        1. No, sorry. Andrew has clearly shown a lack of understanding of sensors, image processing, gain, gamma and noise reduction. In particular by his comments on the use of log in low light show a fundamental lack of understanding of how this works.

          You are basing you opinion of the very vocal view of one person with big holes in his actual knowledge. This is the problem with the internet. Anyone can set themselves up as an expert, but I’m afraid this is the blind leading the blind.

          While it is possible that Sony may be able to fine tune some of the settings of the camera to bring some small improvements, the laws of physics and electronics are at play here. If you add gain you must also add noise. You can then either try to hide the noise which will introduce new artefacts or reduce the NR and then live with more noise. It’s a fine balancing act. Different sensors when paired with different image processors will behave differently. So to expect the A7R and FS5 to look the same is nonsense. To not appreciate that shooting with +10dB gain won’t add noise or other artefacts is nonsense. To think that log is going to give a good image in low light is nonsense.

          The FS5 can and does produce an excellent image when used correctly.

          1. Did some more testing today. I totally understand Andrews point of shooting S-Log in low-light. Of course it is wrong. You explained it in detail and it is the practice I’ve been following for years. But on the FS5 S-Log does not show the tearing at its base ISO of 3’200. So it is the only way to shoot 4K above 1’600 without getting nasty artifacts. Once again, noise is artistically and technically well understood and has been a part of film making since the early days (well as grain). I know it puts a burden on the codec, but if it works in S-Log why shouldn’t it be possible in a standard profile that is much more appropriate not only for low-light work but generally advisable with the high compression rate on the FS5.

      2. Rather than making it personal why do you respond directly to the issues.

        The edge macro-blocking is not just a high ISO noise problem otherwise the FS7 with the same sensor would have it as well. It’s not an 8bit issue as otherwise the 8bit A7R II would have it too. And you can’t blame it on compression either as it’s there on the SDI output. It’s a flaw in the image processing engine.

        I’ll be proved right and I will remind you of that, when it happens!

        1. Oh dear, oh dear, oh dear. No it wouldn’t. They are very different cameras using totally different processing. The sensor is only one small part of the entire signal chain. Just because 2 cameras have the same sensor it does not mean they will look the same or exhibit the same artefacts, image processing is an significant as the sensor.

          1. I wish people would understand that the only purpose ISO serves in a video camera is to allow you to use a light meter to set the exposure. Do people not look at their monitors anymore?

            If you set the camera to 0dB gain and expose a standard gamma correctly, then still at 0dB gain switch to S-log does the image get brighter? The answer is NO it does not. That should be telling you that the actual sensitivity of the camera is NOT changing, even though you are switching between 1000 ISO and 3200 ISO. The ISO rating changes because for normal gammas you would expose white at 90% while for log it ends up at 60%, so if using a light meter you want the light meter to tell you to set the aperture at 3200 ISO to end up with white at 60%.

            This is why I hate the use of ISO in video cameras because it is miss-leading people. The FS5 is actually any more sensitive to light at 3200 ISO in log than it is at 1000 ISO in standard gammas. That’s why when you shoot at 3200 ISO in Log you don’t get any more noise and you don’t get any additional NR artefacts…. because the camera isn’t actually becoming any more sensitive, it’s just a rating change to make sure a light meter would give you the correct log exposure. If people actually went back and used the correct and appropriate terms for video cameras: ie: dB we wouldn’t have all this stupid confusion, it would be obvious that 0dB = 0dB so the sensitivity in standard gammas is actually the same in Log.

    1. Gain is bad, gain is nasty, always has been, always will be. Nothing has changed in the world of video cameras to make gain inconsequential.

  13. Has anyone (who knows what they are talking about) seen any good, I mean really good broadcast quality images out of this camera, HD or otherwise?
    Alister you go on about the Falcon, or should that be the ‘Flacon”? but in this, as nice as it is, you have not resisted the temptation to show what YOU can do rather than what the camera can do. I’ve commented on this previously with you and you replied that “you were after a heavily stylised look”, which is fine but unfortunately does not reveal the baseline of the camera. The nearest I’ve seen is Johnnie Behiri’s ‘Catman”, where real world production can be viewed. (recommend all view this for the narrative anyway).
    You, and all reviewers for that matter, talk about features and not benefits Alister. Having fun ‘grabbing and going’ is irrelevant if you can’t use the material professionally. And if this is touted as a documentary production camera, then it should produce professional broadcast results. You mention the minimum standard of 35Mb/s but this is only for a small percentage of overall production for broadcast.
    So many people are being romanced by 4K as being essential for production, it must be if, as I read today, ‘By the end of 2016 Warner Bros will have released 35 titles on 4K UHD Blu-ray’. Better not miss the boat, eh? Without doubt 4K is perfect for mastering and archiving, at this juncture at least, but you need to have the codec for that to be of any use and the FS5 just doesn’t cut it!
    Sony has sown the wind with their sales bs, and is now having to deal with the resulting storm that is brewing around the world with discussions by intelligent and well informed professional shooting producers. The form factor is indeed brilliant and anyone who has to travel by airplane to shoot, or needs to be unobtrusive on location, has all, I’m sure, appreciated a solution to that part of their job! But gimmicks and shiny badges emblazoned on the body stand for nought if they can only be used with caveats. We don’t want a knock off Swiss Army Knife, just a knife that cuts honestly and cleanly.
    What Sony should take away from all this, is that there truly is a demand for a ‘Ronseal’ camera of this type – something that ‘does what it says on the can’!
    Drew.

    1. So cut to the chase: The Falcon was shot with an out of the box FS5 using a mix of the kit lens and other lenses available to anyone. It was edited and graded on a stock Macbook Pro in an afternoon in my hotel room using Adobe Premiere, software that is available to anyone. What the hell is so unrealistic or unprofessional about that? Do you not appreciate that grading is a typical part of many, many productions these days, especially for many users of large sensor cameras? This is the kind of workflow they are designed for, this is why they have log etc.

      The ability to “grab and Go” is not just about fun, but also about the ability to get shots not possible with larger and bulkier cameras. You can have the best camera in the world, but if it’s too clumsy to get the shot what is the point?

      Not everyone is making movies for the cinema or TV. Corporate video is a massive market, there’s probably more money to be made in corporate for most people than in broadcast. What about YouTube and Vimeo etc. For many people these will be the only places their content will ever be seen, so does every camera need to meet broadcast standards? The FS700 was a huge success, but it never met broadcast standards.

      10 bit, 35Mb/s H264 HD is a permitted standard for 100% of an HD TV production with many broadcasters. The BBC lobbied Sony and others to produce cameras with 35Mb/s H264 for some time as this allows them to use existing SD DVCAM infrastructure. If you bothered to look at the EBU guidelines you will see that 35Mb/s H264 is allowed. But hey, the FS5 can do 50Mb/s HD so it’s comfortably within the permitted specs. And when used correctly the FS5 produces a very nice HD image.

      Buy you’re quite right the FS5 does not cut-it for 4K broadcast or 4K TV. It’s not designed as a main camera for 4K productions, this has never been hidden. But it does make for an excellent B camera for those odd pickups or hard to get shots that can’t be gained with a bulkier camera.

      My FS5 produces very pleasing images when used correctly. Sure I can also make it produce some very nasty images if I use an inappropriate gamma curve or add a ton of gain, but that shouldn’t really be a surprise to anyone.

  14. Would it not be possible to create two more, higher bit rate XAVC modes? Say, a 1080(24/25/30)@100Mbps and a UHD(24/25/30)@200Mbps? The SD U3 standard has a minimum write speed of 240Mbps so that shouldn’t be a bottleneck.
    I understand that doing this is not simply a matter of just changing some values in the firmware code (which, from a technical standpoint, it probably is) but would require Sony to modify the XAVC standard. Besides just helping to alleviate concerns about a single camera, these higher bit rate modes would also help to differentiate the “pro” XAVC-L from the “consumer” XAVC-S.

    1. There already is this… XAVC-I. XAVC-I can go up to 960Mpbs for 4K. It’s 10Mbps for HD 25p.

      I believe the AVC spec can go up to 4:4:4 12-bit, but (unfortunately I suppose) Sony haven’t implemented this in any of their cameras yet (I don’t think).

      1. I seriously doubt that Sony would add XAVC-I capability to the FS5. Only 1080 would be technically possible anyway, the SD U3 write speed isn’t fast enough to support UHD XAVC-I recording.
        A good compromise would be to increase the bit rate of the XAVC-L to maximize the available speed of the SD U3 standard.

      2. The minimum stated write speeds are often impossible to maintain for extended continuous periods without any hesitation. With photography it really doesn’t matter if there is a tiny momentary slow down of the write speed, all that happens is an inconsequential fractional slow down of the write speed for that photo or group of photos. But with video the data coming from the camera is continuous, there is no room for any hesitation no matter how small. So the card must be capable of writing many times faster than the data stream so that if there is a hiccup (perhaps during a refresh cycle or when a bad block is found) the card can catch up virtually instantaneously. Sony are traditionally quite conservative in this area as any hiccup could mean the total loss of the clip being recorded and possibly corruption of the entire card. In a professional product that would not be acceptable. Even with SxS cards Sony only ever write at around 1/3 the of the cards typical write speed. It does sometimes feel like an over safe approach but because of this approach it is thankfully incredibly rare to have any media issues with a Sony camera. Would you want a camera that 99% of the time has fewer artefacts but 1% of the time leaves you with a corrupt card and no footage? Panasonics GH4 does write to SDXC cards at up to 200Mb/s, but it is incredibly fussy about exactly which cards will work and even with the recommended cards speed errors at 200Mb/s are still quite common, especially on cards that have been used a few times. Drop it down to 100Mb/s and you have 100% reliability again.

        1. JVC does h.264 UHD encoding at 150Mbp/s relaiably on SDXC cards (U3)

          If Sony decided to add XAVC-S/L at 150Mbp/s UHD? That would be glorious! That extra 50mbp/s would go a long way to improving the durability of the codec.

          Sony could do a 100Mbp/s 4:2:2 10bit 1080 option on SDXC too. That would also improve it’s HD codec tremendously.

          Sony will never do it though. Unfortunately, Sony has carefully chosen it’s codec limitations to help divide the markets up. Their goal is to keep their long GOP codecs FAR away from their intra-frame codec.

          The last thing Sony wants is to bring the distance between it’s cheap long GOP and it’s high priced Intra codecs performance CLOSER together.

          The performance divide is very intentional. This is a clear marketing $trategy on Sony’s part. Yes, it’s a smart one too.

          CT

          1. At least do your homework before making silly comments about market segmentation. The majority of the PXW line have both the I-frame and Long GoP codecs as standard. X200, FS7, X500, X180, X160……. The FS5 is the only one that only has XAVC-L.

            Besides which the main complaint here is about noise and NR artefacts, the codec is very much a secondary deal. Feed the codec a decent image by using the camera correctly and it produces a very useable image.

          2. “At least do your homework before making silly comments about market segmentation. The majority of the PXW line have both the I-frame and Long GoP codecs as standard. X200, FS7, X500, X180, X160……. The FS5 is the only one that only has XAVC-L.”

            The FS5 is NOT the only Sony camera that has only XAVC-L. Dont you own an X70? Did you shoot your train video in XAVC-L?

            There are 100 different ways the industry uses to split and divide markets. Codec choice, type and bitrates and always a big one. It’s an unspoken thing by the companies but we surely all know this.

            The maximum XAVC-L/S bit rates are unnecessarily low. In UHD, There is a reason for separating Long GOP and Intra codecs so widely. It’s “SD vs. XQD” Sony owns and licenses XQD and is invested in it’s sales. SD cards are not theirs and are of very little help to Sony. You have a choice, stay with SD cards and accept very low long GOP bitrates…or…move to XQD and Long GOP and Intra frame recording and pay Sony a very high price per gigabyte on XQD.

            What Sony doesn’t want with UHD is for you to have a high bit rate Long GOP codec on SD cards that they dont make money on. They will try to say “SD cards cant support anything higher than 100mbps. We keep the bitrate low for your own good”

            Yet JVC and Panny go considerably higher on their SD cards. (Thats OK for them because they dont have an “XQD” in their fleet to cannibalize)

            So no, a 150Mbp/s Long GOP codec would be a very durable one for video on SD cards but that is a bad business decision for Sony. (Even 100mbps, long GOP 10bit 4:2:2 in 1080 would be a bad business decision too)

            I understand the position you are in and I appreciate that you must be careful with your words here. I know people that matter to you are watching all of this very closely.

            I do respect you and this site. You do great work and will continue to do so, I’m sure. I understand your predicament and I don’t blame you for the aggressive tone that you must take on this issue.

            Keep up the good work and I hope that nobody takes this debate too personally, no matter how emotional or frustrating it might get sometimes.

  15. Alister,
    You seem to be taking this personally, and unfortunately it seems that some are making it so. In truth it’s really Sony that should be manning up to these issues.

    I said that your short was very nice when I first brought this subject up and still stand by it, but you can’t second guess how something will be graded or how it will be stylised, which is why base line is essential ‘if’ the point of the piece is about the equipment and not the talent. Your X70 train sequences were much more akin to this and perfectly illustrated the camera, whilst also being enjoyable.

    Seeing as I’m not a ‘tuber, and although I do shoot corporate, it is the documentary business that is our main core. Consequently broadcast standard is what I would want this camera to shoot and I can’t speak for others. It’d be okay if Sony advertised the FS5 as a great little B cam that can’t really handle 4K, and then I could make a clear cut decision that it’s not for me, but the trouble is they make a song and dance about what’s on board. I think the problem seems to be that most feel that it is misrepresented. It seems we all want this form factor, but not with it’s limitations.

    You know, I don’t want pleasing images, I want astonishing images, and more to the point so do our clients. I agree, one is more likely to swing into action with light weight gear but the converse of that is, a crap shot is a crap shot, ok if you’re on the top of Everest or on the moon, where no one else is likely to catch the shot, but at the end of the edit you need to have top notch quality sequences, not ‘they’ll do’ ones.

    I said at the time that the Sony engineers created a little beaut with the FS5, then the sales folks came along and said ‘you can’t put all those features in there that users want – take some features out!’

    However, this would all be resolved if Sony categorically stated what the aim of this camera is to calm expectations, as it is patently clear from the widespread dialogue, that whilst it is ideally suited to some applications, it’s shouldn’t be the choice for others.

    1. You’re not wrong there. Regardless of the fact that their kit can produces “astonishing” images when used well, the links between Sony’s service, marketing and engineering departments are, frankly, a total clusterfuck in my opinion.

      This is fine if one learns to go with their flow (not that one should have to)… but for new comers, it leads to all sorts of complaints and problems, that just reflect badly on them… and thus badly on loyal users too.

  16. I don’t think anyone here will disagree that this camera has been marketed as “Grab and Shoot” and is aimed squarely at event shooters (weddings, concerts, etc), documentary makers. People who currently own a C100 for example.
    Correct me if I’m wrong.. but I don’t think anyone will argue that is exactly what this camera is meant for.

    I shot 81 weddings in the last two years and 4 X 20,000+ attendance size concerts. This camera is aimed at people like me, more than any other type of shooter.

    If a “Grab and Shoot” camera, as Sony markets it, can’t allow me to do 3200 ISO without artifact/image issues/the image going to hell, its one of two things.
    1) Something is wrong with it and it needs to be fixed
    2) I’m completely wrong and its not aimed at people who shoot the type of stuff I shoot.

    I shot 3200 ISO at every wedding and concert I’ve done because thats whats needed in those situations.

    Yes I completely understand what Alister is saying, and adding gain will exponentially increase noise, etc etc. And yes, Alister I agree that the images lots of people have captured from this camera do look great in the situations that they shot in, including your film you posted.

    The point is, this camera is aimed at shooters like myself, who make their living doing events, who cannot control the situations they shoot in, and need to ability to capture a useable image, which means needing to add gain to the image.
    -> Whats the option with this camera? Underexpose so its unusable?
    Or up the ISO to 3200 to get the exposure… but now maybe the NR or something is causing the image to tear, etc etc??

    Yes its not ‘ideal’ for any camera to add gain, I completely agree Alister -> But don’t forget who this camera is aimed at, people who HAVE to add gain to capture what they have been hired to capture. We aren’t using our tools wrong, we using our tools how we have to in the situations we shoot in, and shooting at 3200 ISO when you have to in order to get the shot is NOT using your tools ‘wrong’. “Lazy”… Hilarious. Go shoot a 4-day wedding -> Thursday 12hr, Friday 12hr, Saturday 16hr, Sunday 18hr,….. with no control over locations, schedule, lighting, etc, and see if you need a “Grab and Shoot” tool that can let you do 3200 ISO or not.

    If it can’t do it, its either broken and should be fixed, or they completely missed the mark at delivering for who they intended.

    1. One thing you have omitted though is that really, shooting log and “grab and shoot” generally speaking don’t mix well, and generally speaking are not commonly done together.

      It does seem like these issues are more prevalent in log material. Maybe not exclusively, but it more common, it seems.

      1. And the gammas that are effected most by shooting in 4K are the cine gammas which would lend itself as a profile for less post production.

      2. True -Skipped over Slog -> I almost never use log for weddings. Most of the companies I shoot for don’t have the time to do the legwork afterwards.

        A few of the concert interviews with the bands/stagemanages/lighting people we’ve done in log though (we have 30 mins or so to set those up, so theres time to setup thankfully)

  17. I think that adding gain or higher ISO’s were a problem in the HDV days, but it’s pretty standard for a camera in this day and age to perform fairly well between 3200-4000 iso. Heck the canon’s I’ve been using the past few years, c100 and mark3’s can shoot up to 6400 at events if you need the shot. Yes, there will be added gain which is acceptable, what I would argue and what people are calling out, is the hideous tearing that is happening here. A nice grain throughout the image is far more visually appealing then a mess on the edges of contrast which in many’s opinions ruin the image. I haven’t heard one person deny that grain will start to emerge with the noise floor being raised with increasing ISO, but I think most people here can say this is a problem. I agree that within the limitations of the camera at this time, with lower ISO’s the image looks great and the form factor is fantastic, but there is clearly a problem here. I think the reason many people are speaking up, is that the bar has been set over the past 3-4 years for cameras to achieve great images at higher ISO’s and with other sony models being no exception. I would only expect a new model release to have equal to, or better performance in that area not worse and taking a big step backwards. The images in the a7sii are fantastic looking and not sure why a proper video camera would be crippled by this.

  18. Ive never seen a camera cause such controversy, criticism and disappointment.

    Maybe Sony should have made point to say ‘ALWAYS SHOOT AT BASE ISO’. NEVER USE ANY GAIN.

    The problem is Alister people who do Run N Gun observational shooting i.e documentaries, use gain. Its a plain fact and a little gain should not be causing these issues, especially in HD 1080.

    I never had a problem with the Z1P, EX1/3 etc using 6db+ gain. a little noise maybe but no artefacts etc and they were 25mbps HDV and 35mbps 420 cameras. Real workhorses that built Sony’s reputation for ob doc run n gun cameras.

  19. I have to agree with all these event and doco shooters. I can super over expose a landscape +2 stops and minimize banding too (I’ve done this with the A7 series). However, the problem is it kills your mid tones and can make faces look plastic and you can lose a lot of dynamic range in your highlights. Your clouds are clearly clipped in some of those shots and they’re still some blocky banding evident in the sky when you look. And you’re doing everything short of actively blurring it out and it’s still there. Especially considering the market this camera is aimed for, nobody could possibly overexpose everything just to minimize the noise. None of this argument that you’re making addresses the fact that Sony’s cheaper cameras do a far better job in the same conditions with much less artifacting. You’re absolutely correct, the camera has different processing. It is noisy. It has issues. People can work around these issues, If they know how to, but it’s going to affect the kinds of shots they can get and it directly is against the kinds of shots that the market this camera is made for will be taking. They not taking all the landscapes overexposed, they are in documentary and event situations where they are following subjects from dark scenes, varying exposures, etc. They can’t set up to perfectly over expose the scenes to deal with artifacts that do not show up in a camera $2400 cheaper from the same company.

    The point is people should not have to go through these ridiculous contortions to get a usable image out of it when they don’t have to from other cameras in the line. I don’t have to go to these contortions with all of us Sony cameras why should I have to deal with this one?

    Why can’t you just simply admit that the FS5 is an artifact filled camera compared to Sony’s other large sensor cameras given the same shooting conditions, identically exposed and graded? I don’t understand. I absolutely don’t understand your insistence on blaming the camera operator for what is clearly a camera that has processing problems. It is insulting, it is insulting to Sony’s customers, and it is a horrible representation of Sony to professional sugars worldwide.

    In any case, I’m done arguing. I have work to produce. I know what I’ve seen, I know with the camera can and can’t do, and I have reported as much to others. They can do with that what they wish. Good luck to you all.

    1. If you are over exposing log by 1 or 2 stops and getting plasticky faces, then you workflow is wrong. Every stop above middle grey has exactly the same amount of data, so a face will be recorded with more or less exactly the same amount of information whether correctly exposed or a couple of stops over exposed. If you are having problems with getting the face to look normal then you need to look at your workflow because it’s broken.

      The only “contortion” you have to go through to get a decent image out of the FS5 is to use it as designed. Which is at 0dB. Anything else is a compromise and will result in a compromised image. Why is this so difficult to grasp? Have I missed some marketing where Sony have stated that this camera has some kind of secret sauce that circumvents the effects of adding gain? Have you bothered to look at the samples I shot today. Beautiful, clean images and I didn’t have to go through hoops to get them.

      1. Sure -> How about in Sony Professional Solutions Americas video -> “First Look at What the PXW FS5 Can Do”

        Have a look yourself if you’d like:
        https://www.youtube.com/watch?v=KquhWdL86iE
        17:30 ish

        As someone considering buying it… this is what I get straight from Sony:
        Their filmmaker goes on about how he shot in low-light to emphasize the heat of the furnace…
        Says specifically shooting @ 3200 (which he says the camera was setup for as native)

        “very little noise to it surprisingly”
        “Whoa whoa, thats crazy – even though I didn’t know that, all of our shots turned out just beautiful!”

        And then their filmmaker mentions around 19:00 who this is for…
        “People shooting weddings, documentaries,….”
        “this is exactly the type of camera thats perfect for it”

        3200… very little noise.. and perfect for shooting weddings.
        Thats the FS5 according to Sony’s marketing.

        1. The main point being.. you’re suggesting the reason for the image ‘tearing’ or whatever you want to call it is people adding too much gain and they shouldn’t.

          Sony’s own video specifically says 3200.. has very little noise… beautiful image in low light…and perfect for weddings/documentaries etc.

          Wouldn’t that suggest this camera should be more than capable @ 3200 .. which seems to not be the case?
          Or am I completely crazy?

        2. And the video is there for you to watch so you can judge for yourself. I am not saying that the camera can’t be used at higher than the native ISO, all I am saying is that if you do use it at higher ISO’s or gain levels than native you should expect there to be some artefacts. That’s the normal expectation for ANY video camera.

          3200 ISO is the base ISO for S-log and if exposed correctly the camera does indeed perform very well at 3200 ISO and many testers and user are confirming this.

          BUT Many of the tests being talked about have been using 3200 ISO as the base ISO for the standard gammas. That is completely incorrect, that is the addition of over 10dB of gain, the base ISO for standard gammas is 1000 ISO. When you add that much gain you will introduce image artefacts on any video camera.

          Miss-leading tests and un-realistic expectations from people that should know better have started an unwarranted firestorm. The camera is capable of amazing pictures when used properly. Of course if you want to find flaws in it you can, just as you can with any video camera because none of them are perfect. But when someone starts shouting that “hey this cameras noisy at 3200 ISO” Then I say what do you expect if you add 10dB of gain.

          1. I didn’t want to comment further but with all these numbers being thrown about I want to be clear: Alister, I understand Slog on the FS5 is rated at ISO 3200 but is in fact 0 db gain. You keep telling people their shots in SLog at 3200 is adding gain… when you in fact actually admit as Sony states that this rating is actually ZERO GAIN in slog mode.

            And in fact this 0 db gain still demonstrates the artifacts. My tests were shot in Slog at 0 db gain, so were Joe Simon’s – insofar as Slog goes.

            Yes, I tested increased gain. Yes with increased gain the issues get far worse. But the issues with macroblocking also show in ZERO db gain.

            Zero gain. Let me repeat that again in case it’s lost in all this number discussion. ISO 3200 on the FS5 is 0 db gain, according to Sony (see the other reference in Kyle’s link).

            Forget the cinegammas for a moment. I shot them at 0 db gain too with similar effect (of course this depends on the scene…).

            I don’t know what the A7RII is rated at for native gain but I can assume that its base ISO is in fact 0 db gain and any lower ISO is a pull down. So we shot these cameras at 0 db gain. No extra noise added. Forget the ISO rating because, as you correctly mention, it’s largely irrelevant with a video camera.

            The FS5’s XAVC-L demonstrates blocky, banding artifacts in large areas of color, at ZERO DB GAIN. Again – ZERO DB. Other camera from Sony do not, with Sony’s long GOP XAVC codecs.

            Read that a few times. Again. Again please. ISO 3200 SLog is 0 db gain. No gain. Nothing. ISO 3200 in Slog on the FS5 has been stated publicly by Sony as being zero db gain. So unless someone from Sony was lying, we have been testing at 0 db gain. Against brightly lit skies.

            Yet you insist on making this nonsensical argument that people shooting Slog in ISO 3200 are shooting with +10 db gain, thus it’s noisy and there should be artifacts, etc. So which is it? What gain are they shooting at when shooting slog at 3200 on the FS5? Is it +10 db or +0? Was Sony lying?

            When I was talking overexposure, I was assuming you were saying to shoot ISO 3200 (sorry, zero db gain) slog at 1-2 stops over to minimize noise. This is what I normally do on the A7 cameras, but when I shoot on the FS7 in 10 bit XAVC-I I shoot as you actually recommend, using the scopes and setting white where it should be, or a bit higher depending on taste, etc.

            Of note – my FS7 performs similarly to the FS5 in XAVC-L mode, so this is not confined to the FS5. It looks like it’s the codec. Again, at 0 db gain.

            In any case, please stop chastising people for shooting over 0 db when in fact, shooting Slog at the mythical “ISO 3200” they are actually shooting… 0 db.

            At this point it may seem like I am going in circles but I want to reiterate this point as you have done so too many times for me to count: What am I missing here? You’re telling people to not shoot the camera at 0 db, if you are telling them not to shoot ISO 3200 in slog. Your statements actually contradict themselves at many points and I think even you are getting lost in the numbers.

            ISO 3200 in Slog = Zero db gain.

            So…. let’s back up for a moment and look at actual work. After all, this is why people purchase tools… not to argue over bits and bytes and so forth right?

            Myself, I am a producer and editor primarily. But I have shot extensively with the C300, C100, FS100, FS7, A7S, A7RII, and so on in the last 5 years (never mind the old Betacams and even VHS beasts of old). I’ve been mostly following subjects as part of a feature documentary on Autism that I am producing. I also have covered disaster zones, active shooter training, law enforcement activities, and so forth for the US Government. I have shot in a myriad of situations with difficult practical lighting and changing scenes. You can see some of these shots in my public portfolio and work area of my site, which also includes a (dated) show reel. There’s even more on the US Department of Homeland Security’s TSA YouTube page that I have shot, that I can’t link to on my own company page. I have plenty of work that can speak for itself – mostly of people doing emotive things – and that is a small fraction of what I am actually able to show.

            In all the cameras I have shot with, in scenes ranging from funerals to disasters to schools and everything in between – I have never seen the kind of macroblocking that I have seen in XAVC-L (frankly, on the FS5 and the FS7 both) when shooting at 0 db. Never mind higher gain.

            The camera has issues shooting without artifacts, period. It’s a limitation as you say of a very bandwidth limited CODEC. This is just the nature of the camera… it is what it is.

            We can lay all the science excuses and bitrates and geekery we want in the world down to defend the camera (why? I have no idea, but if you must)… but it negates the art. We must consider WHY people would purchase this camera… and what they will be doing with it.

            They demand a good, artifact minimized image – otherwise why not just shoot on a cheap camcorder? People are buying the FS5 as a “grab n go” “digital motion picture camera” (these are Sony’s own definitions). To do what exactly? To put the camera on sticks with external recorders and scopes and analyze the shot to do all they can to get it clean before shooting? No. They are running after a child in a hallway, or ducking around a corner to film a subject crying in poor lighting conditions, or trying to catch the sunset through disaster debris, or running through dark tunnel after an “active shooter”. They are trying to capture that perfect moment of love in a newlywed’s eye or the agony of a hard day worked out in the cold at a construction site.

            They are, like me, capturing humanity. Thousands of hours of priceless moments are being captured, and the FS5 appealed as a “grab and go”, “motion picture camera” that can capture those moments with a minimum of fuss. It is sad that those shots may be marred by macroblocking that yes, can be cleaned up in post, but is work nonetheless to get rid of. Lack of fuss is certainly not a selling point of this camera, at least when it comes to the images. This is a sad state of affairs, when the body design itself is the anthesis of this: it’s low fuss, high function. It’s beautiful in many respects for this particular large sensor ” grab and go” purpose.

            Real world, emotional documentary and news content is what this camera should excel at. It is literally what it was made for. Are we to expect nothing less than what Sony’s cheaper cameras like the A7RII produce? (Which while noisy, is quite decent as compared to the FS5?)

            Please, please stop making excuses for the technology that does a disservice to the art. People here are trying to do a great job for their clients and jobs, and telling them conflicting information about ISO ratings and db is not helping. Then going off on whatever title people decide call themselves or not (none of which is relevant to this discussion at all) is insensitive at best.

            The FS5 is a very nice camera, actually, but it absolutely has image processing flaws (as does the FS7 in XAVC-L) that people must be made aware of. These flaws will be self evident in many shots when people shoot in environments and scenes their work demands. Not every shot will show it, but when these issues do (macroblocking, color aliasing, smearing on edges) some priceless shots could be damaged or destroyed. Life, after all, is not always at 0 db, perfectly lit, set up on sticks. And life, my friend, is what we are capturing.

      2. Were on earth does it say ‘To get a decent image out of the FS5 is to use it as designed. Which is at 0dB. Anything else is a compromise and will result in a compromised image’ in the manual or hours of presentations.

        Can a documentary camera be used at one ISO especially when your inside a fairly dim room then following someone to the outdoors? What if that fairly dim roomed footage was an integral scene but was unusable because of the artefacts?

        1. Do I really need to respond to this?

          Sure just go out and shoot at whatever gain level you want as it makes no difference whatsoever to the image quality.

          1. Your missing my point or I’m not being clear.

            I give up, I hope Sony don’t with a future firmware fix.

          2. No, I guess you don’t appreciate that adding gain is a compromise and result in images that are compromised.

            For my entire career, over 25 years, adding gain has always been a last resort because the effects of adding gain are totally undesirable. Even when cameras were only the equivalent of 100 ISO we didn’t just chuck in a load of gain to make up for the lack of sensitivity. Today with have cameras with amazing base sensitivities but still everyone seems to want to just add a load of gain and have perfect images.

            The only “fix” for adding gain is not to add it. The camera isn’t broken.

          3. No, I just think you are expecting the laws of physics to change or miracles to happen. If you add gain you add noise. If you try to hide the noise you introduce other artefacts. So you make a choice, shoot without extra gain for a clean image or shoot with gain and accept the INEVITABLE compromises. Sure Sony might be able to fine tune things a bit, I’ve said that from the very beginning, but don’t expect the camera to be noise or artefact free especially when you add gain because that is impossible.

          4. So whats not stated in the manual etc is what your saying if we stick to these we will be fine.

            Slog 2/3 = ISO 3200
            ITU709(800%) = ISO3200
            Standard Gammas = ISO 1000

            I don’t care for 4K 8bit 420 but sticking to above your suggesting will provide best results – artefact free. Im happy with this.

            Would have been more clear if Sony stated this up front.

            Thanks Alister 🙂

  20. +6db or +9db on my EX1R was something that you always wanted to avoid if humanly possible. But that was then. Today, +9db is nothing for even JVC or Canon noise reduction circuits and they are not as sophisticated as Sony’s NR tech. (Yes, I believe that Sony easily has the market cornered on clean sensor readout and downstream image processing).

    I don’t think that anybody here is expecting +9db and up with no image suffering “price” at all. However, as Sony fans, we have become “spoiled” by how good Sony’s gain and image control is today! We expect +9db to look great….why?…because Sony makes +9db look great on ALL of it’s cameras now. (consumer and pro)

    Rumor has it that the Sony A6100 with a 36mp APSC sensor will be out shortly. It will have 4k and a VERY high pixel density based on S35/APSC sensor standards.

    Alister, how much do you want to bet that +9db and higher on this VERY HEAVILY photo-site size disadvantaged camera….will still look amazing?

    People are upset that the FS5’s noise reduction looks “broken” and not functioning the way that it TRADITIONALLY has been with other Sony models. (traditionally=past 3 years)

    I believe that Sony will eventually show you that this FS5 gain issue is not “normal” and they will optimize it to get it back to the performance that it should be. I’ll bet that you will look at the fix and say: “Wow, I guess Sony CAN make +9 or +12db or more look allot better.”

    We are just looking for “normal” Sony high ISO noise problems. The “standard” noise problems that we have come to expect over the years,…not THIS weird pulsating, ragged edge, high gain noise problem that sits all by itself in the Sony fleet right now.

    I have faith in Sony on this.

  21. I’ve owned multiple Sony cameras and I’ve never seen this problem in any of the others. It’s not a codec issue. It’s not a just a high gain issue. And it DOESNT happen in HD. It’s a fault. I can’t believe that anyone would feel this is normal / acceptable:

    1. You have added +10db of gain, of course there will be some artefacts. What you are seeing is the built in NR doing it’s best to overcome the excessive noise without softening the image or smearing the image. You haven’t seen it on any previous camera because this isn’t any previous camera. No previous camera uses this combination of sensor and image processor. Why is everyone ignoring the fact that when you add all that extra gain the increase in visible noise is actually really very small, that there is almost no drop in resolution, and the contrast remains remarkably constant. Everyone is just picking up on one minor edge artefact and ignoring the benefits the NR is bringing everywhere else.

      Why is it different in 4K to HD? That’s easy. If you have a 4K sensor and are delivering an HD image you can use pixel binning to significantly reduce noise, but pixel binning halves the resolution of the image so can’t be used in 4K so other NR methods must be used. This may also explain why you don’t see this artefact with the A7RII which has even more pixels.

      As you have an FS5 try this: Put the camera into 4K, look at the image and switch the gain switch between +6db and +12db. For a very brief moment you will see all the extra noise and grain that the NR is very successfully removing. Would you want all that noise and grain in your images? This edge artefact is a very small price to pay for some really very impressive NR.

      Is it a fault, No, I don’t think so. It’s a side effect of the use of a large amount of extra gain.

      1. I’m not arguing the point that it’s a technical challenge to sensor remove noise, or that it isn’t remarkable how sensitive cameras have become in recent years. However, I am arguing that the way this camera removes noise in UHD introduces visually unacceptable artefacts. I can understand smearing and general softening of the image. But are you saying that you find this ghosting on moving edges and this noticeable blocky disruption of horizontal lines on slight vertical motion acceptable? At just 9db? Recorded externally too? The point of noise reduction is to reduce distracting artefacts without introducing too many new ones. Sony have failed to do that here. They’ve offered a camera that looks like a 4K handheld documentary cam, but it can’t handle many typical run and gun. I know the FS7 is a different camera with higher processing power, but using the same (or a similar) sensor we do not see these artefacts at similar gain levels. Does this not indicate that by fault or by design there is something wrong with this camera?

        1. The FS7 is no low light champion of a camera in fact at its base ISO of 2000 it can be very noisy in the darker tones of an image and the noise is not particularly aesthetically pleasing. You simply could not get away with gain levels that these characters on here expect from the FS5. If the FS5’s unfiltered noise is similar to the FS7’s then it’s no wonder Sony made the decision they did because these people complaining about the NR smearing would be complaining about how noisy the FS5 is when they wind on a gazillion dBs of gain.

          If Sony are at fault it’s not giving the user the option to turn of NR and then the user would be able to make a judgement whether to leave it off or whether they can live with the NR smearing. Both noise and the NR artefacts are unwanted but it’s a compromise to be struck.

          The NR on the FS7 does very little and the real world noise reduction comes from choice of lens, speed boosters and that little thing called lighting. It’s never been cheaper to get high quality lightweight portable video lighting so I take all this “run and gun” no tripod shoot with available light bullshit for what it is, laziness. Why has the video light on the front of the camera been forgotten about? If the subject matter is important enough how can you not bring lighting even if it’s only an on camera light?

          Alister I applaud your indefatigability in trying to educate people.

          1. Agree to everything you have there. Yes variable or NR On/Off would be a good idea and almost certainly be acceptable to everyone.

  22. My take on the FS5 is that it really is a great camera for what it is.

    What do I mean by that. Well I use it as a replacement for my C100. And compared to the C100 it is far superior in HD, 422 10bit. I never had any problem emulating the “Canon Colors” with any Sony cam, just a bit of CC. And it actually has 4K so…

    Is it a baby FS7, NO. And it was never meant to be a 4K Cine cam. The 4K is just a bonus. If I need Cine quality 4K I use my FS7. But in all my tests and I have had the FS5 since early December the images amaze me. I never spotted the ripping problem until I read about it by pixel peepers. And yes then it took me quite a few tries to replicate it. But the only way I can replicate it is very specific and not my way of shooting anyway, so it is almost a non issue in my mind. However I have contacted Sony and got a response from ( I won’t name names here) some senior engineers and they are now looking into what can be done to improve this possible NR problem. And I am sure they will have a statement about it in a while and improvements in FW soon, just like they have on many other Sony Cams.

    In the mean time I am having a great time using this very compact cam for winter outdoor shooting here in NH, mostly HD but some UHD and I love the Slomo. I can say I actually prefer the cache recording to the FS7 continuous. I am so used to the FS700 way and it uses much less media, you only record what you want, you don’t burn through cards like crazy waiting for something to happen. And the media is a lot cheaper than the FS7 media.

    I am sure the dust will settle here soon and shooters will just appreciate this great cam for what it is. Another tool in the quiver.

    For low light I use the A7s, I only have V1 but it is great at that. For serious 4K I use the FS7 as A-cam and the FS5 is fine as B-cam in UHD as well. I rate the FS5 at 1200-1600iso for Log and use 800 or 1000 iso in cinemas. Upping the gain a bit is OK but if it gets real dark I use the A7s.

    Using the FS5 the way I do you will never see the “horrible ripping” you can see blowing pixels up when looking at shaky cam footage shot underexposed with horizontal high contrast lines.

    So I for one am just out there using this innovative cam for what it excels at. And I can’t think of another camera that has all the great features in such a small and ergonomically great package.

    I must say that if you can’t get great images from any of the current available cams it is not the cameras fault. I don’t just mean Sony cams, even iPhones can make great images, and I have used iPhone video in commercials intercut with FS7 footage and the client never noticed. You need to understand your camera, light well and you need to tell a great story. And you need to buy the right gear for what you are doing.

  23. I made a mistake above:
    But in all my tests and I have had the FS7 since early December…

    Should be:
    But in all my tests and I have had the FS5 since early December…

  24. Hi Alister.
    I am considering the purchase of the Sony Fs5 as a main production camera for some of my documentaries that need a low profile camera. Or where we just want to shoot with a lighter, more compact camera. The documentary will be shown on the NRK which requires 50 mb. What resolution settings would you recommend? Is it better to shoot in 4K to have that extra resolution to work it, or will the 4:2:0 codec in 4K be insufficient to meet the 4:2:2 specifications in HD? The NRK will not yet accept 4K as a master so delivery will be XDCAM HD422. Obviously shooting in HD 4:2:2 10 bit will do, but having worked a little with 4K in a DJI Inspire I can see some advantages regarding reframing and extra resolution for stills. I also have access to a PIX 5 recorder but that defeats the small and lightweight somewhat.

    Best Svein Rune Skilnand

    1. For NRK and broadcast I would stick to HD as in HD the camera is 10 bit and much less compressed, unless you do use an external recorder.

  25. Hey Guys,
    I’ve witnessed some of the abnormalities that you’re referring to above. My question for you is this. Should I consider returning my FS5 and wait for a later version to become available? What about a firm wear update? Will this remove the noise and macro blocking?
    Thanks Tom

    1. At 0dB gain the camera works just fine. If you add gain it gets noisy, if you add a lot of gain, you get noise reduction artefacts. This is all normal video camera behaviour.

      If you don’t like the camera, return it for a refund but don’t expect it to get “fixed” because it isn’t broken. There may be some small improvements due to firmware updates, but the firmware isn’t going to alter the basic characteristics of the camera.

      1. WOW. This really blows my mind. Use this camera at 0db and if you don’t like it, tough luck. I feel like I just jumped back in time 8 years. This is 2016 and we are being told this is completely normal behavior to have A LOT of noise at anything over 0db

  26. I noticed something strange when I was trying to grade some of the footage from my new Sony PXW-FS5. When I imported the files into FCPX or Resolve, highlights appeared clipped and unnatural but I couldn’t figure out why. In FCPX’s “Color” module, if I moved the highlight exposure slider up or down I could see added detail in the highlights but as soon as I stopped moving the slider they would disappear and the highlights became ugly and posterized again.
    I decided to try DaVinci Resolve 12 to see if I could get better results. Unfortunately, Resolve also showed this ugly clipping and posterizing. I then decided to look at the file’s “Clip Attributes”. Under “Levels” in the “Clip Attributes” there were three options- “Auto”, “Data Levels 0 – 1023” and “Video Levels 64 – 960”. By default, “Auto” is selected. I decided to try “Data Levels 0 – 1023” instead so I clicked that option and hit “Ok”.
    That did the trick. Now Resolve was showing far more detail in the highlights and the posterization was gone. Unfortunately, I don’t think there is a way to do the same thing in FCPX.
    Evidently, both Resolve 12 and FCPX are not setup by default to correctly render XAVC-L footage and don’t utilize all of the data present in the files. Is this something that you have run into before?

    Here is a screen capture clip where I compare and contrast.

    settings

    1. Edit to my last comment: I just saw your post from 2014 regarding the 109IRE issue and workflow problems. Wow, what a headache.

  27. Hi Alister,

    I’m owning a FS5 and it’s noisy. Therefore I checked all the internet and read all your postings and other informations. Hope Sony will fix it with a firmware update.

    But: You often say, that switching the Camera from Standard Gamma (1000 ISO) to S-LOG (3200 ISO) would not make the image brighter, that should be a proof of your ISO-db-debate … but it’s not correct. The image is much brighter after switching!!

    So there must be an error in the camera or your technical understanding is not perfectly correct in this case.

    Thanks for all the information and the time you’re investing to help the XDCAM-community!

    M.

    1. No it is not brighter but it is flatter. Expose white at 90% in standard gamma at 0dB gain and switch to S-log at 0dB and white drops to approx 75%. Do not confuse low contrast as a brighter image. Look at what the histogram does or better still measure with a waveform.

      It’s really quite simple. If the gain isn’t changing, the actual camera sensitivity isn’t changing, this is very basic video knowledge that people really should understand. The only thing that’s changing is the shape of the gamma curve which alters the viewed contrast. Shadows come up, highlights go down, but the actual sensitivity is the same.

      Ultimately the sensitivity of a video camera is set by the sensor and it’s sensitivity to light. As you cannot change the sensor, you cannot actually change the cameras sensitivity. All you can do is add gain and gain adds noise. Gamma curves are gain curves, so a curve that raises the shadows, like log gamma will not only raise the shadows but also raise the noise in those shadows. This isn’t an increase in sensitivity, it’s an increase in gain (and noise) in the shadows.

      ISO ratings are there for external light meters and do not take into account noise, gamma, dynamic range or any other other vital factors that determine the quality of the final image. There is nothing to stop camera manufacturers nominating almost any ISO they want for a video camera as they can alter the gain of the camera to meet the ISO they want, but that will change the noise levels, possibly significantly.

  28. Okay, thanx for the education part, very helpful. There is – btw – de facto no noise in cinegamma 3 (test five minutes ago with very low light, with O dB, t-stop 1,5, shutter 50, 1080p25), but in S-LOG it’s heavy under these low light conditions.

    I love the camera. Will test the S-LOG with external HD-recording this weekend (Atomos Blade).

    Cheers,
    Michael

  29. Alister,

    What do you think of the data rate? I get a lot macroblocking and some banding with the codec whilst following all your instructions. To an external recorder it looks much better. I can’t help but think that the data rate (100mbs for 4k and 50mb for HD) puts too much stress on the image when you try to squeeze 10bit color, a log gamma curve, etc. I know it can be changed in firmware (a Sony rep said so). If they bumped the rate 200mbs for 4k and 100mbs for HD it would be much improved no?

    Thanks

    Josh

  30. Hi Alister,

    Thanks for all your input on Sony cameras. Your site is a valuable resource.

    I’ve been shooting Sony for around 4 years – the FS700, F55, FS7 and A7SII. All of the cameras have performed brilliantly, yet I do think there’s some issues with the FS5 image performance.

    I completely understand the thing about gain. It’s basic understanding all camera users should know. I don’t think this is the issue, because the same thing happens on the FS7, but only with XAVC-L. I’m convinced there is an issue with the codec/data rate.

    On the FS7, I’ve lit bright highly controlled scenes at 1000 ISO using a Hypergamma, exposed correctly. In many cases, the image looks fine. However sometimes the image looks full of severe macro blocking, smearing, tearing, colour blotches. This seems to happen under these conditions in the scene:

    – Painted walls.
    – Wallpaper.
    – Fine hair.
    – Smoke, clouds, steam, water.
    – Reddish skin complexion.
    – Edges (like the edge of a desk in a close-up).
    – Bright highlights
    – Shadows in fine textures

    If using Slog3, the artefacts are much more severe.

    I’ve seen all the above “issues” in consumer cameras, which is expected. I think some owners are annoyed that they expect better for the asking price. In my opinion, I think XAVC-L doesn’t cut the mustard, and Sony either need to refine the codec and/or bump the data rate. None of these issues happen in XAVC-I (obviously a superior codec).

    The pressing matter is that consumer camera codecs actually perform much better than the FS5’s XAVC-L. The issues are there, but when a GH3 does a better job of handling noise than the FS5’s XAVC-L, even when stressed – it’s a bit of a head scratcher.

    Shooting on XAVC-L can produce great looking footage, however it’s use is limited, especially for the target market for the FS5 – documentary shooters who can’t control their surroundings.

    Do you think a bump in the data rate or change of codec would help improve these issues? Even a paid upgrade to XAVC-I would be welcome.

    Thanks.

    1. 50Mb/s XAVC-L in HD is absolutely fine. This is the same codec as the X200, X500 and is considered as acceptable for HD broadcast. I have not seen any significant codec artefacts from XAVC-L in HD. For years we have been using 50Mb/s mpeg2 for HD broadcast and this was perfectly acceptable, H264/XAVC is a more efficient codec. Perhaps it is possible to raise the bit rate to 100Mb/s but I doubt you would see any difference and we would have to wait another couple of years for Apple to implement the decoder correctly.

      100Mb/s for 4K is pushing the limits. It is not broadcast quality even on the FS7, but it was never intended to be. A small bit rate bump might be possible but then you wouldn’t be able to use the wide range of SDXC cards that you can now. Like the GH4 you would have to be very careful over card choice and even then speed errors would likely occur from time to time possibly resulting in lost footage. I would wan’t to have to explain to a bride and groom why I didn’t get the “I do”.

      My understanding is you won’t ever get XAVC-I in the FS5 as the encoder used can’t do it, my suspicion is that it’s the same encoder chip as the A7. But what we will get of course is the raw update. Then you can bypass the codec altogether.

    2. Thanks Alister,

      That’s a good point regarding HD.

      I’ve seen some very nasty compression artefacts in XAVC-L in 4k, but much less so in HD.

      The FS5 is an interesting camera. Ergonomically it is absolutely top class – and hopefully it can improve these “issues” with the RAW upgrade. Couldn’t imagine plonking a 7Q+ on there though! Hopefully there’ll be a smaller version on the way.

  31. Hi Alister,

    I have issue importing XAVC-L mfx-files into Premiere Pro CC.
    by Draging them into the timeline only the audiotracks gets imported but no videotrack. Do you have suggestions? in the sourcemonitor there is the videotrack visible.
    best Regards

  32. Reading all the wailing and gnashing of teeth here Alister, as I, a small time videographer doing corporate stuff, TV ads, and YouTube videos, and looking at the 8,000 total for the FS5 with 3 year drops and spills, battery, cards, speed booster, and UV flter, and coming from the EX1, which I still have and use, my question is this: Is the image on the FS5 in HD better than the image on the EX1? That would help me a lot. How about the image of the GH4 vs the FS5? I also use the GH4 for 4K interviews. All glass being equal of course, can I beat the EX1 with the FS5?
    Thank you.

    1. I’d suggest renting one and try it for yourself. This is the problem with the internet now. People are prepared to base an important purchase decision on the opinions of people that may have very different criterion to yours. I think the FS5 is better than the EX1 in HD. I have not put the FS5 side by side with a GH4, but it’s not just about image quality, it’s about use-ability.

  33. I made today the test in low light with light area with Cine 4 compare to S-Log 3. Than I compare the files in Resolve. Now everything is clear. Thank you Alister! With Cine 4 I had much more information in the dark area I need. S-Log loosed this information in the dark area. I can`t understand that there is such a big misunderstanding with s-log in the market. Sony does`t help to clarify this.

    Can tell me what Cine you are recommend for which situation?

  34. Thank you very much for the detailed approach.
    A quick question:
    I need to upgrade from my NEX-EA50 to a better option for low-light shooting (events mainly)
    Do you think FS5 isn’t enough for good IQ in iso3200-6400 always used in 1080p mode ?
    (I was initially ready to purchase FS700 but then this spider-camcorder came up)

    1. I think you will find the FS5 to be quite an improvement over the EA50 and much better ergonomically than the clumsy FS700.

  35. Hi Alister, I am a big fan. I’m totally confused by you saying the iso is 1000. From BHPhoto specs:
    Sensitivity ISO Rating: 3200 (S-Log3 Gamma)
    Lux: 2000 lx, 89.9% reflectance
    Video Gamma: T14(3840 x 2160 @ 23.98p mode, 3200K)`

    what am I missing? Please help, I was going to buy this camera but am confused about what I’m reading

    1. What you are missing is that ISO is NOT a measure of the sensitivity of an electronic video camera. It is a number that will allow you to use an external light meter to get the correct exposure for the gamma curve you are using. As the FS5 has many different gamma curves it also has many different ISO’s. In Rec-709 it is 1000 ISO, Cinegamma 640 ISO, S-log 3200 ISO, but these are all the same sensitivity, these are all 0dB gain.

      See: https://www.xdcam-user.com/2016/01/iso-fiasco/

  36. Hi Alister,

    I’m relatively new with the FS5, shot on the FS7 for a while a couple years back. I went out on a shoot today filming some scenery and looking at my footage i’m not very happy with the shadows. I’ve read all the comments and articles but everyone seems to be talking about the UHD 8-Bit performance, I shot everything in 1080, xavc-l, 50mbit today and was expecting it to be noise free.

    My settings were, PP9 (SLOG 3, SGamut3) and ISO 3200 the entire time.
    I had the SLOG assist on the monitor for proper exposure and everything looked good.

    This is what my shadows look like: http://i.imgur.com/SsHdUez.jpg
    I understand the tree is dark, it’s supposed to be dark. I don’t believe it’s supposed to be noisy?

    I’m running firmware 2.00 with the raw upgrade enabled.

    Thank you,

      1. It’s under exposed. Don’t use the gamma assist, it ties you in to 3200 ISO when really you need to rate the camera at approx 1200 – 1600 ISO to get good shadows, especially when shooting HFR.

  37. But in SLOG it won’t let me go any lower than 3200 ISO. Do I always just need to expose the image brighter than what the gamma assist wants me to believe is correct?

  38. Hi Alistar,

    I’ve been doing some testing with 240fps on the Shogun Inferno, and on the ProRes files i’m still getting this odd pink/green pixelated mess around the edges of some objects. This is doing RAW 2k Super Slow Motion out of the FS5 into the Shogun Inferno, ProRes HQ.

    Any idea what is going on here?

    Thank you,

    1. “pink/green pixelated mess around the edges of some objects” that is most certain achromatic aberration from your lenses, nothing to do with the camera.

    2. The HFR modes read the sensor using some form of pixel binning, so there is some loss of quality and an increase in noise. Also the quality of the debayer process in the recorder will effect the mage quality. It’s not unusual to have some very minor artefacts in the slow motion material.

  39. I’m willing to entertain that it could be aberration because they do seem to only be appearing on the edges of overexposed objects, but i’m using the Sony 24-70 G master series lens, so for what I paid for that, it shouldn’t be this bad. They also don’t look like typical lens aberrations, they are digital in form, weird dots and such.

    I’ve filmed plenty of badly exposed stuff on my C100 and it’s never looked like this. Remember this is raw, going into the shogun inferno at prores hq.

    Screenshots:
    Ungraded: http://i.imgur.com/qPEKNv1.jpg
    Graded: http://i.imgur.com/UbXpqAv.jpg
    Zoomed in: http://i.imgur.com/kNisFxm.jpg

    Thank you,

    1. No, it’s not raw. If it’s ProRes then it’s not raw. It is encoded video that has been derived from the cameras raw output (so actually little different from an internal recording, just a different codec). So the quality of the de-bayer process and raw to video conversion of the recorder must be considered in addition to optical and other artefacts. It’s probably a combination of several factors starting with a little bit of CA from the lens, then the on sensor pixel binning to get the frame rate up plus a less than perfect de-bayer process.

Leave a Reply to Josh Swan Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.