Pro video blog…Produced by Philip Johnston DoP/Editor

Investigation title

Today Sony has issued a statement about all the potential picture problems people are discussing on the internet, its probably halted sales and when this happens a company has to act fast.

Sony  “ We are aware of this perceived issue with the PXW-FS5 performance. Sony is investigating this matter as a matter of priority and will work with users to understand the specific issue they are facing. Any actual issue will then be addressed in the most appropriate manner by our engineers”

FS5 hand

Dan Chung “It should be noted that many prominent bloggers who have shot with the camera have not complained of having the same problems. Furthermore Sony Independent Certified Expert (ICE) Alister Chapman published an article questioning why some users are expecting an artifact free image when shooting at 3200 ISO and recording internally in 8-bit?

This has prompted an extremely long chain of comments and debate which you can read if you have the time. I think Alister’s technical analysis of why the issues occur is accurate – especially the part about not shooting S-Log in super low light, although I’m not convinced that the argument that we should not expect better from the camera will go down well with news and documentary shooters who often have no choice but to shoot in very dark conditions.”

HDW “I have not seen any picture problems with the FS5 to date, the only noted picture problem was 2 out of focus shots which were lens related and not the cameras fault. I set my FS5 at 0dB then creep up to 6 or 9dBs if needed. It matches the FS7 in HD 10bit 422 perfectly, I used both of them on a 2 camera shoot late December and the pictures from both cameras could not be better.

If I had seen a problem I would be one of the first to be banging on Sony’s door for a fix but we can only wait for Sony to get back to us with their findings ASAP.”

author

Having been working in the video business since 1988 I have amassed a great amount of knowledge of both the kit and production values over the last 30 years.

11 thoughts on “Sony’s FS5 under investigation

  1. I don’t think that you should explain anything, Philip. I know that you are sincere in your reviews. Not because I trust you regardless, but because you did prove it many times. That said I hold zero confidence on the so-called “official” reviewers based on 1 simple epic fail (the XAVC joke of the X70, when they failed miserably. By then it was I M P O S S I B L E to chant the XAVC beauty out of that camera)
    Then users came out loud and they were all caught with the pants down. What happened with the FS5 was (again) the different processing of the (same) sensor of the FS7. I believe that the trick has been exposed now, no? Not blaming the marketing at Sony’s , they used to be able to get away with the trick for decades. Not anymore, apparently. And again you (Philip) said it loud and clear (quoting “To be fair the Sony FS5 will happily go up to 9dBs with little to no noise in the picture and further if needed but you have to be aware of the limitations, Sony have deliberately not given us a cut down FS7 for obvious reasons”). So you (Philip) are perfectly OK with me. Not the same sentiments for the “official” reviewers (but that’s me, I don’t count anything).

  2. No problems from my own experience. The only thing I’d like is more “pop” in the colours, and I’m working on trying to get that without it looking artificial.
    Every camera has limitations. It’s really unreasonable to shoot at high ISOs. I follow your lead, Phillip; I set the camera to db’s and try to hold the gain to 9 dbs. If I still need more light I get a faster lens. My Canon 1.2 FD with a metabones speed booster will do the trick anywhere I want to film.
    What settings did you use to match the FS5 and 7? Just curious.

    HDW : Set them both to rec709

  3. Hi Mark,

    I’m curious to understand more about your post above. (it’s a little cryptic) What about the XAVC on the X70? I own it and I think it’s a great little camera for the market it was designed for. I was upset about the extremely low 60Mbp/s XAVC and it being the only Sony UHD camera in their entire fleet that was stuck that low. I am extremely happy that Sony has announced 100Mbp/s is officially on the way. Is that what you are referring too?

    On the FS5, what specifically are you upset about? It’s 9db noise performance compared to other Sony Super 35 sensor cameras?

    And for Philip’s quote; “To be fair the Sony FS5 will happily go up to 9dBs with little to no noise in the picture and further if needed but you have to be aware of the limitations, Sony have deliberately not given us a cut down FS7 for obvious reasons”

    I’m curious to know what he means when he says “…for obvious reasons.”

    In my opinion (which means nothing, really) I think that if two cameras share the same sensor with the same signal to noise ratio at the initial read out stage, they “should” perform the same way in 8bit through the entire processing chain. Yes, they “should” be the same 8bit for 8bit all the way up the gain ladder. What processing does the FS7 have in 8bit that the FS5 is lacking? Did Sony “dumb-down” it’s FS5 8bit performance to give extra visual difference between the two? No,…I do NOT know this to be the case, I’m just asking questions to people that might know the true answer.

    It is very possible that the FS5 is limited exactly as Sony wants it to be?. It is clearly Sony’s right to do so. But c’mon, let the FS7 hold the 10bit advantage over the FS5 and let that be main reason why the FS7 is better.

    I wont really know a damn thing for sure until I unpack my FS5 and start testing it against the FS700r, RX10-II and A7S-II. Internet, YouTube and Vimeo videos are a lousy indicator for a $5,000 purchase. (Yes,..I know the A7s-II comparison is not a fair one as they are extremely different cameras)

    I’m really glad Sony is now talking about it in the open. I do applaud and respect them for doing this!

    CT

    HDW : Why would Sony produce the same camera as the FS7 but in a smaller body, that makes no sense, remember Sony have already sacrificed the F5/55 by producing the FS7 to such a high spec though Sony are not known for sacrificing picture quality. Remember the FS7 in most cases is 10bit 422 and only in the FS5 you get 10bit 422 in one flavour of HD. I only need HD from my FS5 so I am sorted.

  4. Cliff :
    https://www.youtube.com/watch?v=rsFJCSPtbxE

    that video explains lots of things. The pro still guys had this discussion (with digital, about low light) around 2009-2010 , it’s time that we do too , and BEFORE we buy new cameras. There is no competition in this industry , but there is the internet, let’s use it.

  5. Philip – “HDW : Why would Sony produce the same camera as the FS7 but in a smaller body, that makes no sense, remember Sony have already sacrificed the F5/55 by producing the FS7 to such a high spec though Sony are not known for sacrificing picture quality. Remember the FS7 in most cases is 10bit 422 and only in the FS5 you get 10bit 422 in one flavour of HD. I only need HD from my FS5 so I am sorted.”

    In my mind, (consumer oriented thinking) the FS7 has all the protection it needs from the FS5 with it’s 4K 60p and it’s 10bit 422 Intra codec. I dont think it’s sales should be threatened by a smaller FS5 cousin that has really great 8bit performance. I dont suspect that people are deciding to buy the FS7 over the FS5 because it’s 8bit video looks better. If that truly is the case, than allow the FS5 to be the best 8bit that it can be at all gain/noise levels. It’s not hurting the FS7 in 8bit.

    Again, I KNOW I’m speaking out of my arse here. I dont own my FS5 yet and I dont know how it really compares to the FS7 in high gain usage. I can only test it against the FS700r in raw/ProRes and that’s a bad test. Testing it against the A7s-II is also extremely unfair.

    Philip, I know you said that 0db was the same on both. Is it the same at +6db or +9db or even….God fobid,…+12db? Yes, I know that it’s TABOO to talk about gain that high on a Pro Super35 camera! (ironically, Sony’s consumer Alpha APSC cameras or VG series Handycams laugh at +9db….funny how that happens 😉

    HDW : My findings are that s35mm as seen on the F5, FS7 and FS5 should laugh at 12dB at least…the larger sensors are far better at containing noise than smaller sensors, usually !

  6. Philip and Cliff : then why a small sensor on pmw400 produces that kind of luminosity compared to a much bigger sensor? It’s all in the processing and the marketing attached to it. Hands down, no questions, case closed , end of discussion.
    Let’s take a look at the marketing with still cameras: they ALL produce the same luminosity at base ISO (between 100 and 200 ISO) , and they keep producing the same luminosity up to 1,600 ISO. More than that and the marketing (and noise reduction) takes place. The marketing with the still cameras is now about REAL added features for the expensive ones (speed , AF and durability, mostly). Now take a look at our cameras: day and night difference at base DB depending on the cost of the camera. Which is still fine , the problem is that in order to sell a decent and honest processing (with no tricks to be clear) and at stratospheric prices they DO darken the cheaper ones. It’s clear and under the sun. This marketing trick needs to stop, like it ended pretty much immediately with still digital cameras. It is simple like that (in my opinion). We just need to push the knowledge with the only tool that we have: the internet. And let’s don’t forget that one thing is to sell features and a totally different thing is to intentionally limit the performance : TOTALLY different.

  7. MARK – “Philip and Cliff : then why a small sensor on pmw400 produces that kind of luminosity compared to a much bigger sensor? It’s all in the processing and the marketing attached to it. Hands down, no questions, case closed , end of discussion.”

    The PMW400 is NOT a single Bayer pattern sensor like the X70. It uses THREE 2/3 inch sensors with 1920×1080 pixels for each color channel. Remember, if you have a single 20 megapixel Bayer pattern sensor, 10 million pixels are green. Red and blue only have 5 million pixels each! Yes,..the x70 has a larger 1 inch-type sensor (it’s not really 1 inch) and it’s jam packed with with 20+ million TINY pixels. The pixel photosites are smaller on the single sensor X70 than the ones on 3 sensor PMW-400. Photosites are often measured in microns. A smarter person than me can do the math for this but the PMW-400 will have considerably larger pixels for each RGB Channel than the X70 does.

    The PMW-400 collects more light for each pixel than the X70 does.

    It does make perfect scientific sense.

  8. Cliff : you will excuse me if I was laughing a little but the position is also kinda serious. Now try to explain what you just said to the still guys who have been enjoying standard luminosity at base ISOs practically from the very start. While we have to keep listening to the most ridiculous justifications and BS that make us spend stratospheric amount of money just for a simple strike of coding in the chips. Remember that Canon did try something similar and was blocked pretty much immediately by a wall of shame (it was the case of the first rebel 300D sporting the same firmware of the 10D but with intentional locks in the code) immediately revealed by the still community. After that miserable attempt no other manufacturer of still cameras ever tried that again. Still cameras are priced and positioned by added features, and not by intentional software limitations. Like we have in the video industry. And let’s don’t forget that even still cameras now sport intentional limitations but only for the video part: like video is clearly considered populated by buyers well used to it. As we are , indeed.
    In response t your theory and justification of the marketing placement and tricks will you explain why the other 3-chips camera also produce laughably dark footage at base db depending on the cost (of the camera)? This “thing” needs to stop, video products must be priced based on added features as well.

  9. Philip : I’d like to share a personal “mea culpa” for not connecting the dots myself about this luminosity thing here (video guys never had doubts before, I didn’t… ). It was many years ago and I was shooting EX1 and my (by then) girlfriend wanted to take stills with her G9: if was a concert of a tenor, in a theater. One spot on the tenor and dim lighting on the orchestra. We get the idea, right? Now my girlfriend took also a short video (10-15 minutes) probably by mistasker, maybe set the g9 on video instead of stills. The g9 shoots 15fps on 1024×768 (4:3). Later I had the EX1 footage in the timeline and my gf asked me to check on her stills and I got her footage on the same timeline of the EX1. The difference in definition and luminosity was jaw-dropping . The g9 was vastly superior (not even close to be honest). I had the same reaction of many of us, basically dismissing it as it was some joke of nature or something. Then everything came back with the first D90. Time to wake up, guys. We’ve been played like puppets , and about the luminosity at base DB then .. oh well… the joke is still going strong.
    /end of rant

  10. Mark: you say “Let’s take a look at the marketing with still cameras: they ALL produce the same luminosity at base ISO (between 100 and 200 ISO) , and they keep producing the same luminosity up to 1,600 ISO.”

    That’s incorrect. There is (by definition) only *ONE* ISO rating for a given sensor which can truly be called “base”. The statement above reads as if you think it’s the lowest the camera can be set to, plus a few others higher? If so, that is not true.

    You have to think about what is happening *AT SENSOR LEVEL* if the input is such as a grey scale. The limit at the dark end will be where an output just becomes perceptible, the limit at the other end will be saturation (more light gives no more output). Note these are intrinsic properties of the SENSOR – pre processing.

    Correlating these against actual light values will give a corresponding value for ISO – and it can only take one value.

    Yes, the processing can be such that the overall camera can be set to varying ISO’s – but go above the base and the results will no longer be optimal.

    That’s intuitively obvious when you raise the ISO – everybody will expect the noise level to rise, yes? Going LOWER than base is much less intuitive, but it corresponds to a lowering of available dynamic range off the sensor. (Whether any lowering of quality of the final image will result is more complicated, and depends on factors such as the dynamic range of the input scene.)

    In a video camera, “base ISO” NORMALLY corresponds to 0dB. Yes, you can use negative gain, but whilst it may slightly reduce noise levels it will come at the expense of highlight handling. (Nowadays the headroom makes it far less obvious on normal pictures. In the 1980’s with tube cameras, any negative gain whatsoever could mean that you’d never get a volt of output – no matter how wide the iris was opened. You’d get saturation clipping before the normal white clipper at 1 volt.)

    And for a given sensor technology, base ISO will largely depend on individual photosite size. The only caveat is when a high count sensor gets downscaled to a lower resolution. (There is an effective “averaging” such that behaviour is more equivalent to a sensor of the same area, but fewer, bigger photosites.)

    Regarding why some 3 chip cameras (in your opinion) produce “laughably dark footage at base db”, the answer is in how this optimal DR *OFF THE SENSOR* gets subsequently processed. As simply as possible, if the basic DR is high, it can *either* be processed to give good highlight rolloff and such that would normally be considered ” a nice picture” *or* processed to give a “brighter” picture. (Hence a higher ISO be presented to the user.) In a pro camera such as Alexa, F55 etc producing a “nice picture” is of prime importance. Which is why a far more expensive camera may seem to a user to present the need to be used at a lower ISO setting than a cheaper model. Yes, it’s counter intuitive, and took me a long time to even begin to get my head around it – but think of what is happening at sensor level, and how that basic signal is subsequently processed.

  11. David Heath: nah, in order to have a decent low light they reserve the software with no darkening tricks only to the 10K and up ones.
    So they do it because of the money. And they did it pretty well so far. Remember that even the day before a recall (to avoid lawsuits) there will be internet posts defending them. Then the next day they get the recall. Funny, uh? It happened with the 5D recalling the mirror falling down because of a bad adhesive. and the day before you had all those idiots saying that there was no problem with the camera. Are you still laughing? I am.

Leave a Reply

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