Other: XAVC-I vs. ProResHQ

I never started the aggression or rudeness I came here and with respect eager to share my findings in the hope it would help others. I came here to share my information but Doug started this whole nonsense with his aggressive condescension then others joined in. But it's OK if Dougie does it, right? I should just sit and take it without responding, right? Wrong. I'm too old and knowledgeable to be spoken down to by a proven half-wit.

Anyway, I hadn't thought about the FS7 for 4 years and I haven't even picked a camera up for 4 years either but at least I've got to the bottom of its poor performance. I'm going to concentrate on enjoying the FX30 and leave this forum to you all, you do deserve each other. The knowledgeable people have moved on and the dotards remain so there's no point being here. No wonder this place is full of tumbleweeds.

Nicely stated.
If you hadn't gone back and deleted and edited all your posts from that first thread a couple of weeks ago I'd be able to show you how I treated you with nothing but respect until you got pissed that someone had the nerve to question yuor outlandish statements and ask for evidence to back them up. You don't like being questioned. You want to be treated as if your word is gospel and we shoudl all just accept it.

That's not how it works on a forum. You need to be able back up your claims. You couldn't. Now that you know you are wrong, you want to blame me, Sony, and whoever else you see as causing your confusion and incorrect conclusions. And we are also evil for not correcting you fast enough or in the manner you think we should have. You're a real piece of work.

Sincerely, your friend and chief dotard, Dougie ( a founding member of the dribbling geriatric ignorati club)
 
Last edited:
No, when you over-moderate is when readers flee, and this has happened here. It will settle itself without intervention.

!00% agree with Tom. It will settle itself. We are all supposedly adults here and should surely be able to realise getting vitriolic and twisted over what is at the bottom line just a technical discussion solves nothing. Nor does killing the conversation. If someone abuses someone fine, take some kind of remedial action. But don't block free speech. I feel confident in what I said because it's been based on the empirical evidence I've seen while working with XAVC since its inception. I've made my points on how I think XAVC is a good, solid codec when treated and processed correctly. Those who learn to process XAVC correctly will discover this. Many shooters and editors have discovered this. Some unfortunately, for whatever reason, never do. What's the old saying? "You can take a horse to water, but you cannot make it drink if it doesn't want to."

Chris Young
 
I knew when I wrote the above I'd get a sideswipe blaming my process but what you ignore is the fact that a the process I followed is an entirely accepted and advertised workflow by Sony. Go watch Alister Chapman videos on SLog3 from way back, Sony and their camera ambassadors regularly advertise the fact that SLog3 is designed to over-expose in this way but no one ever explained how this would affect the image due to the way Sony distributes the 10 bits along the SLog3 curve. I bet everyone here who had an FS7 thought they were getting 10bits all the time. If it was all so easy one of you would've explained this earlier but none of you did so let's not pretend it was easy to get to this point of understanding. I may be wrong but this is the only logical conclusion I can make given the data available.

Show me anyone who said back when the FS7 was a new camera that exposing two stops over would negatively affect the compression quality, no one did. We all knew that rating the camera at 500 ISO would reduce dynamic range that's all. Anyone who now turns up and says otherwise is a historical revisionist.

The hardware had an unpublished limitation with the advertised workflow. I made no mistake, my workflow is completely blameless because Sony did not make these crucial pieces of information available and they're still not available. My workflow exposed these flaws because I was simply not just transforming SLog3 into REC709 I was creating highly graded and crafted content for clients.

I have noticed more macroblocking in SLog3 with the FX30 compared to HLG3 and S-Cinetone. I bet this is due to poorer use of the 10 bits in SLog3 than the other gamma curves or a wider spread between bits at certain points on the curve. You'll never notice if your only ambition is to turn SLog3 content into REC709, my ambition is to go much further and create highly stylised content which does push the image further which is why I use HLG3.

If you are not a Sony engineer and you are presented with footage containing ugly compression artefacts do you blame the camera's internal processing or do you think it's the compression algorithm? I defy anyone to have concluded it was anything other than the compression causing these issues. CYVideo stating that XAVC can't have changed made me re-examine things, if XAVC is fixed then there must be some other explanation, that explanation is contained in the previous post. Thank you CYVideo for a crucial piece of information that forced a rethink and probably a better explanation for the FS7's visual artefacts.

10 bit log will have artifacts when pushing too much in grading. Better to use 12 bit log raw or 16 bit linear raw, or the best, arri 12 bit (14 bit internal) log uncompressed raw, or 12 bit uncompressed dng on external recorders/monitors like o7q+ atomos inferno. I see significant difference when grading 12-16 bit raw vs 8-10 bit log. I have to be very careful to be light handed to not break the 8/10 bit log footage.

I don't have hands on experience on FX30 and FS7, so I cannot do the tests by myself. Just some guess work. FX30 is a full frame camera, FS7 is a S35 camera. For the same scene at the same lighting conditions, full frame cameras have less noise than S35 cameras, so even using the same codec, noise occupies less bandwidth in FX30 than in FS7. As more useful information is recorded in the footage, there is more overhead room for grading. Some one with FX9 can test in S35 mode, comparing FS7, do heavy grading, I bet they will behave similarly, but less satisfying than ff modes.
 
Last edited:
I don't have hands on experience on FX30 and FS7, so I cannot do the tests by myself. Just some guess work. FX30 is a full frame camera, FS7 is a S35 camera.

Hand on experience is actually very important.
For example, in this case it would inform you
that the FX30 is actually a super 35 camera….
perhaps you are referring to the FX3?
 
Hand on experience is actually very important.
For example, in this case it would inform you
that the FX30 is actually a super 35 camera….
perhaps you are referring to the FX3?

Sorry that I assume all FX cameras are full frame.
Maybe FX30 as a S35 camera has a better sensor than FS7 in terms of low noise at high ISO?
Another possibility is that FX line is said to use Venice color science. Maybe FX30 has Venice style color processing units so that it preserves color information better than FS7?

Anyways, I think for heavy grading, raw or RBG 4444 are the better choices.

For my practice, I am torn by the choice of using EI s log 3 or custom modes with in camera adjustments. Log footage when applying S log 3 to Rec 709 A LUT generally gets good results in Resolve 18. But if the scene is not contrasty, in camera custom modes seem generating better results sometimes. For S log 2 or 3, 10 bit 422 already loses much information, the benefit of doing heavy grading seems not much. Probably in camera adjustments without S log is the better way to go, especially for ENG projects? But on the other hand, it is very nice to have the high light protection of the S logs. It is difficult for me to make in camera adjustments to protect high light as good as S logs.
 
Last edited:
Sorry that I assume all FX cameras are full frame.
Maybe FX30 as a S35 camera has a better sensor than FS7 in terms of low noise at high ISO?
Another possibility is that FX line is said to use Venice color science. Maybe FX30 has Venice style color processing units so that it preserves color information better than FS7?

Anyways, I think for heavy grading, raw or RBG 4444 are the better choices.

For my practice, I am torn by the choice of using EI s log 3 or custom modes with in camera adjustments. Log footage when applying S log 3 to Rec 709 A LUT generally gets good results in Resolve 18. But if the scene is not contrasty, in camera custom modes seem generating better results sometimes. For S log 2 or 3, 10 bit 422 already loses much information, the benefit of doing heavy grading seems not much. Probably in camera adjustments without S log is the better way to go, especially for ENG projects? But on the other hand, it is very nice to have the high light protection of the S logs. It is difficult for me to make in camera adjustments to protect high light as good as S logs.

When you talk about the pros and cons of in-camera adjustments, which camera are you speaking of because they are not all the same?

When you talk about "heavy grading" of S-LOG, are you talking about normal grading with properly exposed footage? Would you call that "heavy grading". Or are you talking about footage with serious exposure, WB, or other problems -- or perhaps taking good footage and giving a very extreme creative look?
Just trying to understand exactly what you mean.
 
Last edited:
When you talk about the pros and cons of in-camera adjustments, which camera are you speaking of because they are not all the same?

When you talk about "heavy grading" of S-LOG, are you talking about normal grading with properly exposed footage? Would you call that "heavy grading". Or are you talking about footage with serious exposure, WB, or other problems -- or perhaps taking good footage and giving a very extreme creative look?
Just trying to understand exactly what you mean.

F5 specifically.

For S log 3, although Sony officially claims that IRE 65% is the best to protect the high light, I set my Zebra to 90%. When I record, I typically check to see the zebra showing up, then turn down to not see Zebra. Although EI ISO 2000 is the official value, I set to ISO 1000 most of the time, unless when it is low light, I have to increase the ISO values. In Resolve, I typically apply only the LUT of S Log3 to Rec709A, do a little bit of adjustment on contrast, saturation, denoise, ect. This I consider as light grading. If I want to do heavy grading, I will do something more on color wheel, exposure, etc. I like punchy images. And a lot of big budget films' styles are actually high contrast high saturation, without losing much highlight at the same time, like John Wick 4, and earlier episodes. I assume that OP's talking about the stylish grading is like this. Of course, there are other stylish styles mimicking or utilizing Alexa's high dr cool color science.
 
Last edited:
F5 specifically.

Yes, you have plenty of in-camera adjustments avaialble on the F5 to create a wide variety of looks if you choose to avoid using LOG. On other cameras, such as the FX6, your choices are far more limited due to having virtually no paint menus. That's why I was wondering what camera.
 
F5 specifically.

For S log 3, although Sony officially claims that IRE 65% is the best to protect the high light, I set my Zebra to 90%. When I record, I typically check to see the zebra showing up, then turn down to not see Zebra. Although EI ISO 2000 is the official value, I set to ISO 1000 most of the time, unless when it is low light, I have to increase the ISO values. In Resolve, I typically apply only the LUT of S Log3 to Rec709A, do a little bit of adjustment on contrast, saturation, denoise, ect. This I consider as light grading. If I want to do heavy grading, I will do something more on color wheel, exposure, etc. I like punchy images. And a lot of big budget films' styles are actually high contrast high saturation, without losing much highlight at the same time, like John Wick 4, and earlier episodes. I assume that OP's talking about the stylish grading is like this. Of course, there are other stylish styles mimicking or utilizing Alexa's high dr cool color science.

I agree with your definition of light and heavy grading. But I strongly disagree with your exposure methodology. You are WAAAAY over exposing S-LOG and the results will absolutely suffer. I've done serious side-by-side testing in 5 IRE increemnts between 60 and 95. The sweet spot is around 70-75 on bright reflected whites.

Sony actually claims 60 or 61 is the correct target, but I believe they purposely put that low so they can claim higher native sensitivity for the cameras. It also assumes someone will be grading for "cinema" levels in post rather than broadcast. For cinema-style grading output 60 is doable, but even then it is right on the cliff of being underexposed and if you miss the target you have no wiggle room. 70-75 works great and is a safe level for both broadcast and cinema levels of grading. That's my 2 cents, even though nobody asked! -)
 
I agree with your definition of light and heavy grading. But I strongly disagree with your exposure methodology. You are WAAAAY over exposing S-LOG and the results will absolutely suffer. I've done serious side-by-side testing in 5 IRE increemnts between 60 and 95. The sweet spot is around 70-75 on bright reflected whites.

Sony actually claims 60 or 61 is the correct target, but I believe they purposely put that low so they can claim higher native sensitivity for the cameras. It also assumes someone will be grading for "cinema" levels in post rather than broadcast. For cinema-style grading output 60 is doable, but even then it is right on the cliff of being underexposed and if you miss the target you have no wiggle room. 70-75 works great and is a safe level for both broadcast and cinema levels of grading. That's my 2 cents, even though nobody asked! -)

Thank you very much for the advice, Doug. I will try your suggestions next time. Best regards,
 
It took me a long time to figure out why recent feature films mostly are on the dark side. At first, I thought there are aesthetic reasons, the so called "hell" style, dark and cool color. Now I am enlightened that this may be due to technical reasons. Most feature films in the last 5 years were shot on Alexa series cameras. Probably most DPs shot in 12 bit ProRes 4444 Log C. In order to be able to grade for large theatre screens, they have to use low IRE percent, therefore, underexpose the scene. If shot on raw, typically DPs go ETTR, if the scene looks dark, it is easy to lift half a stop without destroying the footage.
 
Unfortunately there are a lot of knuckleheads who think under-exposure equals "cinematic".

When I watch great feature films from 1980s to early 2000s, they are very sharp, well-lit, have good color. I don't know why now most people consider "filmic" or "cinematic" as being underexposed, desaturated, and soft.
 
Last edited:
Throughout film history there have been many films that might be considered underexposed (most famously The Godfather films) or desaturated (and some only soft because of the type of film used or how it was handled in post).

Besides the occasional look piece, most movies in Hollywood today are pretty saturated and sharp, but many do lean more towards the darker side because low-key lighting has always been more attractive.
 
Back
Top