That's the one instance where I might find Technicolor's style useful - in very low-key scenes with bright lights within the frame. The lifted blacks would give a little more latitude in the shadows.Originally Posted by liverpool 1
OTOH, you can lift the blacks in post in any other style, too...
My S-curve isn't ideal in all circumstances, either, in some cases with low-contrast but well-lit subjects it will tend to crush the mid-tones together a little - we have only got 8 bits, you know!
Here is a test I did today, had some spare time...
I shot a greyscale off my monitor with different settings on the camera, and plotted how the different picture style settings record it (note that the original was on a monitor screen, thus not comparable to the photographed scales, but those can be compared):
Here, I have lifted the blacks of all others to the same level as Technicolor style, so they are more comparable:
Here's a quick test I've been able to work on yesterday:
I should have recorded another sequence showing the Neutral profile too, but I totally forgot. Anyway, here's the three main "flat" profiles compared.
You should try lifting the blacks of both Marvel and S-curve to make them comparable...
I just wanted to show the difference between the profiles in the same conditions.
I'll try to work a bit better on it later this week, to make a deeper comparison.
Just for some infos: I'm working on a video that shows Marvel Cine/Janke's S-Curve/Technicolor comparison, and a original&color corrected with some random videos shot at night using Janke's preset. The later video are shaky, being shot while I was driving, but shows how good it can look in available/low-light (with a f/1.8 lens). I'll upload it probably tonight.
I've done some tests and it seems that using certain types of film grain to dither an image with lots of banding can cause unwanted flickering when uploaded to Youtube/Vimeo.
Look at the sky in the third shot:
http://vimeo.com/27671934
Perhaps because it was not contrasty enough when I shot it? I wonder if add noise would perform better...
I got directed here by drapeama, from dvxuser, and I must say: we've been on a very close path!!
exactly these motivations (excessive noise using CineStyle, difficulty grading with Marvels Cine and other flat picture styles, which usually lead me to grey-ish skin tones and clay-looking people) are what have pushed me to create my Similaar suite of Flaat picture styles:
http://www.similaar.com/foto/flaat-p...les/index.html
except I started with some advantage over Janke: I realized that my monitor's output is not linear, and I used my camera's light meter to create a reasonably accurate DR test chart
http://www.similaar.com/foto/flaat-p...est-chart.html
using that, I learned that this is what Technicolor Cinestyle does:
this is Marvels Cine:
and this is Neutral with contrast=-4:
watching this, I felt I could bring an improvement, so these are Flaat 1 through 4:
(also: all of that is for the luminance channel; on the color side of things, with my skin tone tests, I learned that, not surprisingly, I like Portrait best, so my Flaat styles are based on that one)
lots more info in my Flaat presentation page, here:
http://www.similaar.com/foto/flaat-p...les/index.html
I can think of one other scenario-lighting condition when Technicolor would be the best choice. A high contrast scene with important highlite detail you wish to preserve by underexposing 1-2 stops.
I'd rather use the camera's highlight priority (ISO D+) setting instead - it provides one stop more latitude in the highlights.
(Maybe underexpose a bit if the highlights are more important than the shadows, and the contrast is too high for Neutral at its lowest contrast.)
I've got some tests of precisely that with lots of different picture styles, including CineStyle and my suite of Flaat picture styles
http://www.similaar.com/foto/flaat-p...d-tests-1.html
http://www.similaar.com/foto/flaat-p...d-tests-2.html
why has my previous post not appeared yet? I got a message that a moderator had to approve it, but I guess that should be done by now, right?
it had lots of nice graphs, and links to these two pages:
http://www.similaar.com/foto/flaat-p...les/index.html
http://vimeo.com/34943260
Maybe it simply didn't went through the Spam-O-matic filter...due to the links and graphs maybe, as you're a new user here. You can always pm a mod and link your DVX user account.
It's approved now. Thanks for your very interesting and informative webpages!
(PS: My own "S-curve" style will flatten the mid-tones, so it's definitely not ideal for portraits... but it does bring out a bit more in highlights and shadows...)
I have not been able to revive any previously-clipped highlights: I can increase contrast there, but the clipping point won't move even a third of a stop furter, no matter how aggressive the curve is
in the shadows, Flaat_4 brings up more than one stop of information that's fully clipped in Technicolor CineStyle and Marvels Cine (that obviously comes at a cost: it will be noisy, and it may create macroblocking, banding, etc, in the rest of the image; but if you need as much extra DR as you can get, it works)
where can I find your .pf2 file? I could run it on my DR test chart and post here the shape of the light response
edit: as I said, I think we started out on a very similar path
I would really like to see a comparison with the same graphs, thanks!
My S-curve style is here: http://www.saunalahti.fi/animato/temp/JEStyle.pf2
Of course, what is clipped, is clipped... But if you underexpose just a little bit, and also use the D+ setting, you can do a lot, really.
yes, but if you're going to underexpose then the curve you want is totally different: instead of bringing down the highlights so they're no lonker clipped (a loosing battle) you want to bring up midtones and shadows and compress the highlights without clipping them (like flaat_3 and flaat_4 do, if compared with Marvels Cine and CineStyle)
I've worked with film for over 35 years, and there, the gamma curve is S-shaped, compressing both shadows and highlights. Highlights are never "clipped" on film, unless severely over-exposed - but you may lose them in printing, if you need to bring out detail in the shadows... (Underexpose, and you lose all shadow detail!)
Paradoxically, I'd like to have the same kind of response from my T2i...
ok, so I downloaded that pf2 file, and used it to develop my RAW stills of the DR test chart, and build the synthetic 1/3 chart
I did so both with contrast=0 (the default of the picture style) and contrast=-4 (I thought that might be the correct option, but from the results I guess that's not the case)
contrast=0
contrast=-4
I guess the good one is the first one, but I'm afraid I don't like even that one too much: weird and wiggling light responses that will be very difficult to grade (at least with my very limited skills)
my guess is that you developed this ps by looking at a very specific image, and it's great for that one, but not really so much as a general purpose option
my guess is that you'll like Flaat_1 best
but Flaat_2 is not far either
my post with the results is waiting for moderation again
in the meantime, I'll add this: a way to check that my graphs are correct would be:
find a place where you have absolute control on lighting, and light evenly some surface (a white wall will do)
put camera with Magic Lantern on tripod, pointing at that surface
go to M mode, set white balance, load picture style
set aperture at max
go to live view
set slow speed that will make the light meter tell you you're just about clipping (100% IRE)
start taking speed up (faster), 1/3 stop by 1/3 stop, and take note of ML light meter value on each step
graph said set of numbers
Thanks - your test confirmed what I already knew - my S-curve style isn't the best...
I knew it compressed the midtones, but until I saw your curve, I didn't know how much...
More or less. I was trying to improve highlights and shadows in landscape/cityscape scenes - it is definitely not good for portraiture where the midtones are important!
I have loaded Flaat2 and Flaat3 into my 550D - they both look good! Thanks!
I'll have to do the same here. So far I think I'm fine with Marvell when shooting video. Good compromise. I'll definitively look into these Flaat profiles soon!
Quick update:
Check out this short film shot by a fellow dvxuser forum member, entirely using my Flaat_2 picture style, on a 60D:
http://www.similaar.com/foto/flaat-p...d-tests-3.html
Some shots are just lovely. Even some that are usually quite difficult for any camera (I suspected highlight rolloff would be nice, but had never pushed the camera so far, so his results in this respect surprised even me).
And even more, when asked about how much post there was in the footage, he said: "I ran a color correction pass through it but it was dead easy. I played with the curves (crushed the lift a bit) and added a slight grain for a retro feel." So just what I was aiming for: clean footage with nice skin tones and easy to work with in post.
I'd take that over "all hues are correct", any day. But in any case: the next version of the Flaat suite will include two sets of picture styles, one based on Portrait (for people like me that have trouble getting nice skin tones back after shooting with picture styles based on Neutral) and one based on Neutral (for people with better color grading skills, and that would rather not have these color shifts). I will release that next version when I have the picture styles for the 5D3 too (I think the massively improved high ISO and the better codec could mean greatly improved DR in video: something like Flaat_4 could no longer be just a specialty option).
My personal feeling is that the blacks were lifted a bit too much, not even the deepest shadows were really black, but a dark grey. (I do have a calibrated monitor.)
I would have dropped the blacks just a bit more in post.
Posting Permissions
Reply With Quote
