Shooting flat, anyone?

C

chervas

Active member
Here is a very interested blog by Editman about the limitations of the AF when trying to shoot flat, http://theeditman.com/blogg/2012/02/04/LookingAtLoglikeRecordingWithTheAF101.aspx.
Has anyone had any success shooting flat, or log if you prefer, and then grading without the image falling apart as the article indicates?
I should add that I am having a little trouble following the argumentative logic in the blog. I am hopping The Editman jumps in here and helps me along. How do the 8 bit limitations are an impediment to shoot flat, but the same 8 bit limitations become mostly irrelevant after shooting through a scene recipe that is not flat, flatten through software and then grade it? Is the camera unable to capture enough detail and DR under a flat scene file? Are you saying that the image capture while shooting a clean scene file like yours contains already the DR and detail that one strives for when shooting flat?
Thanks
 
I'm not sure I follow the blog either. It seems like he's saying to shoot contrasty, flatten in post, then color correct. How does that make things any better?
 
Shooting flat with 8-bit and AVCHD works within the limits of the system but it breaks easily. Most shooters want a flat image recorded so that they can get it right in post and have the greatest dynamic range from the camera. You can either shoot with the image as intended to look or shoot with the intention to fix it in post. If you shoot it very flat with the internal AVCHD 8-bit system and try to stretch the curve it will break much sooner than if you shoot it contrasty and then make it more flat in post. That's because you don't stretch the curve in making it flatter.
 
This is by no means a good procedure to test any camera for this topic. The internal scene files actually bake the effects into the recording, so crushing your image and blowing out your highlights on an 8-bit 4:2:0 system will inevitably end in tears, especially if what you are looking for is a wider dynamic range. I wouldn't take any theory like this seriously unless it's backed up by rigorous and exhaustive industry accepted camera tests. Also, the screen grabs AND video need to be full res and available for download for total comparison (before and after) in each test. Look at the Zacuto shoot out for example. There are so many varied situations that can determine actual dynamic range.
 
I think those pictures show that this approach has merit. I have been trying to light for contrast lately. I like a little latitude in both directions but I tend to prefer things a little flat. CineV is unruly for me. I prefer HDNorm and Norm2 to Cine D so I don't get that noise in the shadows. I have been testing for a dark film and have gotten great results lately but honestly it's more a function of controlling the light to manage contrast. I also find lighting brighter the areas for exposure and stopping down f2.8 - 4 at ISO 400 while flagging the light off gets very little noise in the shadows.
 
If I understand correctly, the article is saying you can have a middle ground between flat 8 bit and flat 10 bit? I just don't understand how this is possible. In post you can flatten the image a lot but you're never gonna get that detail back once it's baked in or out by the camera. I don't know, maybe I'm just not comprehending this correctly. But it isn't this a bit like taking rusty water, adding red dye to make it look like Kool Aid, and then trying to change it into lemonade? Mind you it would have been easier to make lemonade in the first place. Maybe this is a bad example...but that just shows how well I understand the concept.
 
IMHO, the gamma curves on the AF100 aren't quite intense enough to emulate a compression
system like SLog (and the companion post LUT). Cine-V with a low knee is probably the closest.
Jacking the master pedestal up is a poor choice as you're chucking bits you can never get back.
Plus, jacking the master pedestal can significantly raise the noise in low-light parts of the image.
If anything, I'd take the master pedestal slightly down (-2 -> -5 ).

I have tried playing with DRS at setting 1 ( as 2 and 3 scare the **** out of me). The problem is,
I can't figure out exactly the curve DRS is applying and the appropriate LUT in post is a mystery
(since it's continuously adjusting based on the contrast of the image). If you could freeze the DRS
curve .. then I think that might work in post. But, DRS is still a bummer because it raises darks
and schmooshes lights. If one could defeat the dark raise, then I think you could get a curve
very close to an SLog curve.

jeff
 
First when i film i shoot all skin at 50-55 IRE,
If its natural light and its in the frame I don't let more then the light closest to the buld hit a 100.

Then in premiere its important that the Neat video reduce noise is the first layer.
Then its the levels, setting midtone to 1,75-2.0 streches the dynamic range by x1,75 or x2.0
That drains out the color in the image so I ad some with 37% added saturation.
 
Igelkott Film said:
I have made a HDR preset in Premiere pro that I really like.

Download video

Its:
Reduce Noise - Defult filter and sharpen
Levels mid 1.75
Saturation 137
Click to expand...

What settings did you use on the AF100?

To be honest ... what I'm seeing is the level and saturation was raised .. which raised the noise of the darker sections.
The noise filter did a reasonable job .. but the rock walls still have the "crawling with ants" look.

If you were doing a "flat" gamma curve on the AF100, your initial scene wouldn't have the amount of contrast that's shown.

With a "flat" curve, everything looks "schmooshed" .. so the lights wouldn't be quite as light and the darks would
pretty much stay put.

So you get a really "washed out" look in post. With the right LUT, it expands the dynamic range. So .. I would question
your HDR technique (no offense).

jeff
 
Last edited:
jmarkham said:
What settings did you use on the AF100?

To be honest ... what I'm seeing is the level and saturation was raised .. which raised the noise of the darker sections.
The noise filter did a reasonable job .. but the rock walls still have the "crawling with ants" look.

If you were doing a "flat" gamma curve on the AF100, your initial scene wouldn't have the amount of contrast that's shown.

With a "flat" curve, everything looks "schmooshed" .. so the lights wouldn't be quite as light and the darks would
pretty much stay put.

So you get a really "washed out" look in post. With the right LUT, it expands the dynamic range. So .. I would question
your HDR technique (no offense).

jeff
Click to expand...

No offense taken, I'm just sharing :)

V detail -2
Detail coring -1
Gamma b.press
Knee mid
Matrix norm
Others standard.
 
Unlike the Sony EX series, the tweakability on this camera is limited. Im a tinkering soul, so this is very missed. I've never been able to find a flatter setting than Cine D. Period. I shoot with Cine D with DRS-1. Although, I have never put the camera through the gauntlet of a full blown dynamic range and noise test. I will try this when I have some spare time and upload.
 
Ok, let me see if I can answer some of my own question based on the posted responses.
The camera’s technical limitations make impossible to create and shoot a flat scene file.
These limitations are:
- Weak gamma curves cannot resolve enough detail under a flat scene file.
-DRS function cannot be defeated and it will try to compensate a flat setting by increasing master pedestal.
Not sure that not being able to shot flat is intrinsic to 8 bit cameras.
Shooting uncompress will indeed lends grater latitude while grading, as Ilgelkoff proved, but the limitations will persist and flat shooting is again not possible.
Shooting a clean scene file with contrast, such Editman’s, does seem to render nice grading results. I just not sure why the image needs to go through the flatten post stage if indeed it does not stretches the S curve, and therefore, it does not increases DR or gamma detail. I can see it from creative process, as if you were creating a starting point to your grading by muting the colors of the canvas. Editman’s settings and practices do seem to clean chromatic aberrations and it produces a noise free image. Editman, I would like to know your step-by-step process. Are you using Colorista?

Please feel free to ad and or correct my summations.
It does seem to be a common tread running through my ownership experience with this camera. At all times you must be aware of protecting this camera from itself.
 
chervas:

I believe the word 'impossible' is a bit strong in this case. It is a 'flat' profile, just not a heavy logarithmic exposure placement. While I can appreciate trying to squeeze every bit of detail into a frame for post, I would not assume that selecting anything other than Cine D gives more latitude unless you put this theory through a proper camera test in a controlled environment. This calls for a serious testing session with results for everyone to see.
 
I shoot fairly flat and color correct in post with no issues. I use Barry’s setting and it works for me.
The AF-100 doesn’t have S-log or whatever Canon C300 has so its not an option. I like what I end up with and I don’t get to hung up on the tech stuff. Try a few settings out and edit it yourself then you be the judge. To much talk on limitations and pushing the codec until it breaks. Unless your into really heavy post work I don't understand what the fuss is about. If you really want s-log type settings and large dynamic range pony up about three times what you pay for the AF-100 and you got what you're looking for.
I do want those same things but I got what I paid for and I’m good with that.

Cheers!
 
The flattest setting that works is when you shoot in Cine D and dial down chroma to about -5. You'll get some noise and if you stretch the curve gradients can start to break up but it works pretty well giving you a little bit more room to grade. I think this camera preserves some information in the highs and lows that can be brought forward even in a more contrasty gamma like Cine V and the upside to shoot in Cine V is the low amount of noise and when setting master pedal to the right level gradients holds much better.

I use PP to grade most of the time or sometimes AE with the MB plugin. Try to use MB when shooting with my clean settings in an outdoor scene and put in a fill filter and shoulder filter. You can make the image flat looking and bring detail back in both highs and lows without running into much noise or gradient break up.
 
jmarkham said:
I have tried playing with DRS at setting 1 ( as 2 and 3 scare the **** out of me). The problem is,
I can't figure out exactly the curve DRS is applying and the appropriate LUT in post is a mystery
(since it's continuously adjusting based on the contrast of the image). If you could freeze the DRS
curve .. then I think that might work in post. But, DRS is still a bummer because it raises darks
and schmooshes lights. If one could defeat the dark raise, then I think you could get a curve
very close to an SLog curve.
jeff
Click to expand...

Jeff, I agree with you. What most people don't realize is that DRS overrides whatever gamma is selected. DRS is its own gamma, the top end being much like Film-Rec, but the mid-blacks are dynamically lifted and change shot by shot. With an HPX3700 or 2700, you can select Fim-Rec gamma and then a dynamic range of between 200% and 600%. Think of 600% being similar to DRS 3. Then, there is a black stretch control also expressed in percentages to lift the mid-blacks, but also adds noise.

Problem is, a P2 Varicam has a 32-bit DSP and the palmcorders just don't have that kind of horsepower, in addition to fighting the dynamic aspect of DRS, which is really impossible to light for because your fill light values are changing shot by shot. So, that leaves Cine D, which is not as flat as Film-Rec or Filmlike 3 in 2/3" Panasonic cameras. The Canon C300 has C-Log, which is similar to Film-Rec, except 800% dynamic range, which expresses the maximum range of the sensor, like the P2 Varicams at 600%. The EX1/R/EX3 have more film gamma options and a bit flatter than Cine D, but nothing like Sony's S-Log.

The takeaway is that when using DRS, you are no longer using the gamma you thought you were and you cannot consistently light a scene with DRS when shooting coverage.
 
Dino said:
The takeaway is that when using DRS, you are no longer using the gamma you thought you were and you cannot consistently light a scene with DRS when shooting coverage.
Click to expand...

I don't fault Panasonic for the DRS option .. I think I know what they were trying to do. Unfortunately, their compromise "easy to use" solution makes
it very difficult (some would say impossible) to expand in post .. and can raise noise in lower luminance parts of the image to unacceptable
levels.

For those that suggest reducing the chroma value in the scene setting, what effect to you attribute to the gamma curve (which is
luminance based)?

IMHO, in a 4:2:0 color space, I would hesitate to significantly reduce chroma values as you're just throwing the
bits down the drain.

jeff
 
You must log in or register to reply here.
Back
Top