Indie Filmmaker Color Grading Suite Setup

Home Forums Help and Support Indie Filmmaker Color Grading Suite Setup

Viewing 13 posts - 16 through 28 (of 28 total)
  • Author
    Posts
  • #4364

    Florian Höch
    Administrator
    • Offline

    Right, ok, so does that mean the Tone Curve on the 3D LUT tab is not meant to be the ‘source’ gamma, but the gamma you want to apply to the source?

    It defines the source and thus overall tone response when the 3D LUT is applied.

    let’s say that if I grade two versions of my film to look exactly the same through both a 2.2 gamma and 2.4 gamma calibration LUT, and output both to DCI to display in a theater at DCI’s 2.6 gamma, which will turn out darker, 2.2 or 2.4 version?

    The higher the gamma, the darker the result. But note that BT.1886, even though it uses a nominal (technical) 2.4 exponent, is often closer visually overall to a 2.2 gamma curve (with more open shadows though) due to taking into account the display black level.

    This last response is confusing. As we’ve discussed previously, the 1D calibration (vcgt) is global and affects all output equally.

    Yes, it affects all output, but doesn’t override the tone response enforced by the 3D LUT – it’s the other way around (same as in ICC color managed applications, the tone curve of the source profile is what determines the result).

    #4368

    Shane Taylor
    Participant
    • Offline

    The higher the gamma, the darker the result. But note that BT.1886, even though it uses a nominal (technical) 2.4 exponent, is often closer visually overall to a 2.2 gamma curve (with more open shadows though) due to taking into account the display black level.

    Agreed on the BT.1886 comment as I remember reading something to this affect before.

    However, in respect to choosing a Tone Curve on the LUT tab in DisplayCAL, let’s look at a hypothetical situation. I’m thinking this because it came up when I spent months setting up/calibrating/aligning my 5.1 surround audio monitoring system when creating my soundtrack for the film. Different physics, but similar issues.

    For example, let’s say I have some standard Rec 709 Gamma 2.2 footage and I have two 3D calibration LUTs, both with Rec 709 color, but one with a gamma of 1.8 and one with 2.6 (different enough to be noticeable). I grade the footage twice using each LUT in turn, ensuring that ultimately the footage looks exactly the same on my calibrated monitor. Now, when using the 1.8 LUT,  I would have seen an overall lighter image, so I would have lowered those midtone values in that footage file to create my desired ‘look’. Similarly, when using the 2.6 LUT, I would have seen the exact same original footage on my monitor as much darker, and would have raised the midtone values in the footage to create the exact same desired ‘look’.  So, now my two graded files have different midtone values (both different from the original), with the file graded with the 1.8 LUT having lower midtone values, and the file graded with the 2.6 LUT having higher midtone values. When I play each file at the theater, the file graded using the 1.8 LUT will appear darker than the one graded using the 2.6 LUT (because the data in the midtones in the file are lower).

    The choice of LUT changes your view of the footage and therefore affects your grade represented by the data in the file. Each file’s data will then be interpreted differently downstream. Does that make sense? Only if the file contains metadata that informs the renderer of the expected gamma will the gamma 2.6 version be darker than the 1.8 version, and I don’t know if that is standard practice or not. This metadata doesn’t exist in audio files, but it is important in determining how your graded footage (or audio) will be interpreted downstream. Is this thinking correct as it pertains to video?

    • This reply was modified 7 years, 6 months ago by Shane Taylor.
    #4375

    Florian Höch
    Administrator
    • Offline

    When I play each file at the theater, the file graded using the 1.8 LUT will appear darker than the one graded using the 2.6 LUT (because the data in the midtones in the file are lower).

    Yes (I re-read your question again, I missed that you said “after grading them to look identical”).

    Only if the file contains metadata that informs the renderer of the expected gamma will the gamma 2.6 version be darker than the 1.8 version, and I don’t know if that is standard practice or not.

    Depends on the encoding format and container. For h264, for example, the metadata consists of color primaries, transfer characteristics, and matrix coefficients. Transfer characteristics seem often to be ignored by decoding software when played back though (which is probably intended, because usually the display used for playback is assumed to have a suitable transfer characteristic already).

    #4396

    Shane Taylor
    Participant
    • Offline

    Cool. Thanks Florian.

    #4429

    Shane Taylor
    Participant
    • Offline

    Florian,

    So, I finally got a chance to Verify my calibration. I’ve attached the HTML report. Would you please have a look and let me know what you think?

    Many thanks! If this looks OK, I will start grading!

    Cheers,

    Shane

    Attachments:
    You must be logged in to view attached files.
    #4431

    Florian Höch
    Administrator
    • Offline

    Looks alright. It seems the white point has drifted a bit since the calibration was done though.

    #4432

    WolfPeace
    Participant
    • Offline

    Could this be due to display not being warmed up(turned on for 30min +)?

    #4433

    Shane Taylor
    Participant
    • Offline

    Monitor was warmed up for a couple of hours prior.  I did see that shift too. I’ll run it again and may do through te whole process over again to check for consistency of my devices.  6200 sounded familiar though, even though I’m pretty certain I put in the correct x,y coordinates in the calibration.   BT .1886 doesn’t change it does it?

    #4434

    Shane Taylor
    Participant
    • Offline

    By “alright” Florian, other than the WP shift, anything else that could be better or improved? Would you consider this setup workable?

    #4437

    Shane Taylor
    Participant
    • Offline

    OK, so I ran the full calibration/profile/LUT creation sequence again, and once again verified it. I’ve attached the file. The white point is on, and the delta-Es are better, but the gamma drifts above 80%. I don’t get it. The only thing I changed from the last run, was that I turned off the CIECAM stuff on the Profile Tab (and used the ‘video’ extended verification chart). Any ideas? I’ve also uploaded the settings (ICM) file so you can see my settings (right?) I’m going to run it again tomorrow. Sleep now.

    Cheers,

    Shane

    Attachments:
    You must be logged in to view attached files.
    #4441

    Shane Taylor
    Participant
    • Offline

    Florian, just to clarify, before I run a Verification, I need to ensure that the Calibration data from the current profile of interest, is loaded in the vcgt, right? When I went straight to Create 3DLUT after profiling, I don’t recall seeing the option to install the profile, and later, after immediately running the Verification, found that the profile I just created wasn’t installed in Windows (and I assume, the calibration data not loaded in the vcgt). That would affect my results, right?

    #4442

    Shane Taylor
    Participant
    • Offline

    So, I ran the full calibration again, start to finish, including adjusting monitor controls interactively, and using 2667 patches this time while profiling, and made sure that profile was loaded in Windows and the vcgt loaded before running the Verification, this time with 343 (Large test chart – video) patches. I’ve attached the new report here. It seems OK (to me), white points slightly off, and a few gamma points off above 95%. Is the gamma issue common, and should I be overly concerned about it? Not sure I’m going to get any better than this, unless you have some recommendations to changes to my settings. Plus, given that I’m using an i1 Pro and and i1 Display 2, who knows if this is even close to being correct, right? It’s almost like ‘why bother’.

    Any final advice would be appreciated. Thanks for your patience and assistance, Florian. Great program. Plan to contribute to your efforts when the dust settles.

    Cheers,

    Shane

    Attachments:
    You must be logged in to view attached files.
    #4453

    Florian Höch
    Administrator
    • Offline

    made sure that profile was loaded in Windows and the vcgt loaded before running the Verification

    This isn’t strictly necessary for verification (but doesn’t hurt either), it’s only important that the profile that’s being verified is selected under “Settings”.

    I’ve attached the new report here. It seems OK (to me), white points slightly off, and a few gamma points off above 95%.

    Looks good. White point off by less than one delta E is negligible.

    and a few gamma points off above 95%. Is the gamma issue common, and should I be overly concerned about it?

    Not really. It is normal for gamma to fluctuate above 90%, because a very small (imperceptible) change in luminance will already effect a comparatively large swing in gamma (btw, I’m going to remove the 1% grayscale near-black and near-white increments from the testchart in the next update, they don’t really serve any purpose. Originally I had meant these to be a tool to check for clipping, but there are better ways to do that, and fluctuating gamma seems to be a common cause of needless concern for many people that look at the graphs).

    Not sure I’m going to get any better than this

    It’s a really good result already.

    Thanks for your patience and assistance, Florian. Great program. Plan to contribute to your efforts when the dust settles.

    Thanks, contributions are always welcome and appreciated 🙂

Viewing 13 posts - 16 through 28 (of 28 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS