Are Flat Picture Styles Snake Oil?

Started by Deny, December 14, 2016, 06:00:28 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Deny

I see there are several different opinions on whether or not shoot flat when not recording RAW, some say it's mandatory others say it's a bad idea.

A quick disclaimer first, I own a 70D (ML still in beta), a 6D (ML RAW resolution lower than 1080p) and a 5D Mark IV (ML... God knows when). So currently my best choice is still shooting without ML, especially when doing multi camera stuff.

Now I have tried grading ALL-I 1080p footage shot with flat profiles and I must say, with less than stellar results. I've tested ML 1080p on my 6D and it's awesome, but I can only record 1080p 6 seconds at a time so it's a no go.

I have purchased both James Miller's DeLUTs profiles and EOSHD C-Log profiles, have given them a try and while the results with DeLUT 2 have been good, I decided to turn on the Live View histogram in my 5D4. When I did, what I've found, correct me if I'm wrong, is that what flat profiles actually do is just raise the black level, lower the white level, or both. Whether anything above or below those points is clipped or compressed to me is irrelevant as in practice what's happening is I'm losing color resolution, again, correct me if I'm wrong.

Anyone tried this with their cameras? Turn on histogram, select picture style, look where the black and white clipping points are, compare with other picture styles? By the way I'm starting to understand the preference by some well known DSLR shooters for the "prolost" and Marvels picture sytles, because they don't compress any levels and thus let one use the full color resolution.

Maybe I'm missing something, but if we have 256 levels for each channel which is already too little, why on Earth would we want to reduce that number? No wonder people talk about banding when using flat picture styles.

Frank7D

I can't give you a technical explanation, but my understanding is that a "flat" picture style will compress the image tonal range before it hits the codec, thus allowing a lower contrast look that would not be possible merely by manipulating the compressed video in post. It certainly seems to capture a different range of information (although not more information) so depending on the situation it could be a valid tool. Whenever I shoot H264 I use a flat picture style and am happy with the results.

Andy600

It's not so much the Picture Style but the 8bit internal H.264 encoding that makes log picture styles ultimately detrimental to latitude in the final image (when compared to simple flat settings aka ProLost settings).

The only potential benefit is through raising the black level to just above the point where the H.264 algorithm gets really bad i.e. deep shadow detail. You can actually get away with relatively fewer code values in deep shadows without there being much impact on the image - this is how most modern log functions are designed (Log-C, BMD Film etc). Raising the black level in a PicStyle emulates this.

Technically speaking you can compress any camera's DR into an 8bit space but anything over ~6stops gets extremely lossy. There are so few code values written that the image will just fall to pieces when you try expanding it i.e. the more grayscale range squashed into 8bits = the bigger that gaps/holes will appear between cv's when expanded = bad banding/contouring.

Cinestyle was initially developed solely to make inter-cutting with log footage easier, not to maintain greater DR (see the Picture Styles thread for some interesting findings). The same image can be achieved after the fact using flat settings and a lut so there really is little point to log picture styles. Picture styles that emulate Canon Log but recorded to H.264 is just asking for trouble - it will seriously impair highlight information. It might not be noticeable in well lit, normal DR shots but throw some big contrasts or gradients in there and it will soon show it's weakness. I predict it to be much worse than Cinestyle.

I actually did an experiment a while back (still need to write it up) where I produced perceptual log images that matched Cinestyle, Marvels etc from a ProLost source image outside of the camera using simple 1D luts. This maintained the most DR and latitude (latitude is important) and would still allow inter-cutting with log footage in a much less destructive way through inverting the lut i.e. Cinestyle to Rec709 = visible banding in Rec709 space whereas Rec709 (flat settings) to Cinestyle = little or no banding in Rec709 space - because it is already (perceptually) encoded as Rec709.

So to answer your question. Flat Picture Styles and Log Picture Styles are 2 different things. Relative to 8bit H.264 internal encoding, a Flat PS can be useful but a Log PS will typically throw away more information when there is only just enough to start with. Snake Oil - probably yes! BUT when there are more than 10bits to play with things get flipped and a 10bit log encoding is a much better way to store the information.

Colorist working with Davinci Resolve, Baselight, Nuke, After Effects & Premier Pro. Occasional Sunday afternoon DOP. Developer of Cinelog-C Colorspace Management and LUTs - www.cinelogdcp.com

Danne

QuotePicture styles that emulate Canon Log but recorded to H.264 is just asking for trouble - it will seriously impair highlight information. It might not be noticeable in well lit, normal DR shots but throw some big contrasts or gradients in there and it will soon show it's weakness. I predict it to be much worse than Cinestyle.
Yes, I tested and worse. Clipping(correct term?) in shadows crushing finer detail.

QuoteI actually did an experiment a while back (still need to write it up) where I produced perceptual log images that matched Cinestyle, Marvels etc from a ProLost source image outside of the camera using simple 1D luts. This maintained the most DR and latitude (latitude is important) and would still allow inter-cutting with log footage in a much less destructive way through inverting the lut i.e. Cinestyle to Rec709 = visible banding in Rec709 space whereas Rec709 (flat settings) to Cinestyle = little or no banding in Rec709 space - because it is already (perceptually) encoded as Rec709.
Interesting.

Levas

The enemy here is the h.264 compression.
h.264 is made to compress, it will look for areas with small color and brightness difference and compress it in a way that it gets the same color/brightness values and thus saving data(and losing detail).
So what do you think happens when you feed h.264 a flat, low contrast image...it will find lot's of areas with detail that has such small difference in color/brightness values, that it can compress the hell out of it  :P

Maybe you should experiment in filming the same scene with two different picture styles, one flat, desaturated profile and another profile, like 'landscape'.
I think the non flat picture profiles gives you bigger files, even with All-i compression.

That said, you can't color correct much in a non flat picture profile...


Andy600

Quote from: Levas on December 15, 2016, 10:41:52 AM
The enemy here is the h.264 compression.
h.264 is made to compress, it will look for areas with small color and brightness difference and compress it in a way that it gets the same color/brightness values and thus saving data(and losing detail).
So what do you think happens when you feed h.264 a flat, low contrast image...it will find lot's of areas with detail that has such small difference in color/brightness values, that it can compress the hell out of it  :P

Maybe you should experiment in filming the same scene with two different picture styles, one flat, desaturated profile and another profile, like 'landscape'.
I think the non flat picture profiles gives you bigger files, even with All-i compression.

That said, you can't color correct much in a non flat picture profile...




Makes sense. Maintaining contrast and saturation should produce a superior encoded H.264 image. I need to test the matching functions I tried between ProLost and log Pic Styles but substitute ProLost for default Picture Styles settings (with sharpening off) to see if perceivable latitude increases - I guess grayscale might take a hit but color will be better.
Colorist working with Davinci Resolve, Baselight, Nuke, After Effects & Premier Pro. Occasional Sunday afternoon DOP. Developer of Cinelog-C Colorspace Management and LUTs - www.cinelogdcp.com

Deny

Really interesting replies, thanks guys :)

As you've probably guessed, I'm now where some of you were some time ago, that is, testing profiles myself. Last night it was VisionColor's turn and while the footage straight out of the camera looks good, it also produces some weird color casts that make grading a bit difficult, of course further testing and confirmation are needed.

On a side note, I wonder if I could get away with log PS when using the 5D4 4k MJPEG option, being 4:2:2 and all, but that's not going to do much good if I have to edit and grade in multi camera mode along with footage from the 6D and/or 70D.

For now the PSs that give me the best results are Marvels, Prolost and DeLUTs 2, but I still want to conduct more tests before I settle on a workflow.

Now pardon me if this is too much of a n00b question, but isn't in-camera PS contrast simply applying an S-curve? If so, setting contrast to the minimum value would mean a less pronounced S-curve and thus a more even distribution of values along the Y-axis (output), meaning the data sent to the codec would have better, not worse detail. No?

SittingNow

This is not an issue on cams using the mjpg codec. I've been testing the DeLuts Clog profiles on the 5D4, and the results are great. I need to run a few more tests, then I'll post some results.

dfort

Quote from: Andy600 on December 15, 2016, 10:27:47 AM
...a Flat PS can be useful but a Log PS will typically throw away more information when there is only just enough to start with. Snake Oil - probably yes! BUT when there are more than 10bits to play with things get flipped and a 10bit log encoding is a much better way to store the information.

If you mean Snake Oil as in, "a remedy for all diseases" then yeah, I totally agree.

I think we should define what a "Log" picture style is because technically all picture styles are log. What I think would be an accurate definition is a picture style that is designed to capture as much of the camera sensor's dynamic range as possible without regard to how the image appears aesthetically and allowing as much latitude as possible for color grading in post production. The log settings in professional cameras that record 8-bit, like the Canon C100, are hardwired into the firmware and probably don't work the same as downloadable "log" picture styles. What is interesting with the C100 is that most of the posts I've seen on the subject show that users seem to prefer shooting Wide DR Gamma which is a "flat" picture style rather than the Canon Log Gamma setting.

We should also define what a "flat" picture style is. It is basically a gamma (contrast) adjustment of the Rec. 709 mode.

When I first learned photography, my college professor had us shoot what seemed like endless tests of classmates holding Kodak gray cards in every conceivable lighting situation. Then he would have us plot the H&D curve of our negatives.

H&D stands for Hurter and Driffield who were the 18th century photographic scientists that discovered an interesting phenomena which is that when plotted on a graph using density as the y axis and the log of exposure on the x axis instead of getting a straight line they got an S-shaped curve. There were three parts to the curve, in the shadows there's the toe, the majority of the curve is more or less a straight line and the angle of the slope is called the gamma and finally the highlights fall on the shoulder where the curve flattens out and adding exposure has no effect--well sometimes adding a lot more exposure will reverse that part of the curve and is known as solarization. Ever seen a black sun in a photo? When shooting black and white we learned to expose so that important shadow detail is above the base + fog level and develop for the highlights so that important highlight detail isn't compressed into the flat shoulder part of the curve.

Well that was a whole semester's class in one paragraph. I like to see pictures and graphs so here's one that illustrates what an H&D curve looks like:



The base of film isn't completely transparent and the maximum density doesn't block all light so the usable area between base plus fog density and D-max is what we've got to work with when shooting film. Likewise, with 8-bit digital we have between 0-255 bits to play with and in 10-bit we've got 0-1023. That's a big difference but--most color displays are only 8-bit. Wait, you say, but my monitor is marketed as a 24-bit display! Yes, it is an RBG display, 8-bits per channel, so the maximum range is still just 0-255 per channel or -- 256 shades of red, green, and blue, for a total of 224 or 16,777,216 color variations. A 10-bit display can theoretically show up to 1.07 billion color variations. I say "theoretically" because I've worked with both 8-bit and 10-bit displays and 99% of the time I couldn't tell the difference. About the only time that it really matters on a display is when you have subtle color gradations like in soft clouds then you can clearly see "banding" in 8-bit displays.

Back to the film lesson, we don't look at negatives we look at a print, either on paper or on film projected on a screen. Prints are usually higher in contrast than negatives and print emulsions show brighter whites and darker blacks than negative emulsions. Properly exposing a negative involves measuring light and making calculations while make a print is more of an art. Likewise when grading log and flat picture styles we're looking at it aesthetically, not scientifically. The more you have to work with in your original the better be it a film negative or a flat or log (preferably 10-bit and up).

Can you mix flat, log, H.264, 4:2:2, 4:2:0, etc.? Sure, it is done all the time. Will it match in a multi-camera setting? Depends--it is often hard to match cameras that are the same make and model. It is best if you have the chance to do some testing before the shoot. Which takes us back to that college photography class--shoot someone holding a gray card (or color chart) in every lighting situation you are likely going to encounter.

[EDIT] Sorry, got a little wordy there and forgot to point out this post about picture styles. Seems like most of this was worked out in 2011 and there hasn't really been any breakthrough flat picture styles for DSLR's since then.

http://philipbloom.net/blog/pictureprofiles/