[Greg Sage]

After messing with colormunki and other items, I see that colormunki’s generated profile is added to the profiles in System Prefs, so it replaces those in the list. Seems like the control I have there is in the beginning of the process what I’m calibrating for. Choices are:

ITU-Rec. BT709 (Video)
NTSC (Video)
PAL SECAM (Video)

I wasn’t quite sure between the first two, but being in the US, I chose NTSC. Is that the right choice?

From there, it does everything for me except brightness and contrast, then spits out a profile.

On the positive side, it has an ambient light sensor (which can vary quite a bit in this room) that can measure every few minutes and adjust, so I’m assuming that addresses the gamma issue (although I could take a light and dark profile separately and switch manually if that would be better.

On the negative side, it just bypassed all the other settings on the tv which makes me wonder if they were optimized first.

For instance, the tv itself has a gamma setting, 11 point white balance, and quite a few other controls that were not addressed. Should I be doing some process to dial these items in first to get me closer, and THEN using colormunki to do all of it’s overrides in it’s custom profile?

Also, other folks have suggested that I view in 4:4:4 while doing visually critical work, and I’ve read that this tv can handle the input, but I don’t see anythin gin either the existing color profiles, colormunki setup, or anywhere that indicates the ability to do this. Where and how would that be enabled? (Or should it be… since viewers won’t see it that way?)

[Greg Sage]

Looking into some of the other items you mentioned.

In the meantime, there are lots of color profiles listed (and cut/paste doesn’t work within that box), but the color options I see that mention 709 are:

HD 709-A
Rec.709 Gamma 2.4
Rec.709 PQ

Which bring up one other slight point of confusion. I’ve got decent control over the tv’s calibration including specific gamma settings in 0.2 increments, but not clear if that should be matched to the card’s output?

In other words, I’m in a room that receives plenty of indirect sunlight (and no artificial light), so I’m assuming that should be accounted for in the calibration of the TV, but is it the TV’s setting ONLY that should reflect that, or also the output of the card to match?

[Greg Sage]

OK, then I basically need to boil that down into practical steps to get up and running. Unfortunately, I’ve wasted great time and effort over the years by not quite nailing down each of the following steps, but I have an opportunity now with the new setup to finally get it right (to the extent the hardware allows)

As I see it, I have to define settings for:

1) AE workspace:
I’m guessing this should be rec709 and 16 bit, right? I do have a practical question here: Since I’m used to working in rec709 8 bit, with the 16bit color just double ram or scratch disk use or something like that, or will it effectively double everything like render time per frame? Also, is there a specific VERSION of rec709 that I should be aware of as I go down the chain (I notice I have several versions for video card options.) Also, if the video card is set to output lower color bit, is there something I need to do here to let AE know that the 16bit rate is only INTERNAL, and not to be shown?

2) AE render:
I’ve been told before that I should output at full workspace quality, then compress with Media Encoder, so I’m assuming that’s still the case. Assuming 1080P, and whatever is worked out for working color space above, what would those other details be to set the output render?

3) Media Encoder output:
I’d assume that’s just whatever YT’s upload requirements are…. though I do remember doing some experiments and getting higher quality by submitting higher quality than they requested, 2 pass encryption, etc. Any insights here for optimum YT playback definitely welcomed.

4) Checking the render:
It’s been a while, but if I remember correctly, the YT version ends up being sRGB, right? I think the way I was handling this was always causing the YT version to have much more contrast than my working version, and it essentially ruined every video. I also remember something about QT player ignoring gamma settings, and not being accurate reflection of YT playback whereas VNC was closer to YT playback. Whatever the right answer is here, I need to make sure my YT render looks like the workspace version as much as possible without the rec709 -> sRGB contrast shift, and that I’m viewing it in a player that accurately reflects the YT viewer experience.

5) Video card settings:
Again, I see multiple versions of rec709, gamma 2.2 or 2.4 outputs. I’m in a room with plenty of indirect sunlight, but it’s not what I’d call “bright”. It’s all indirect sunlight, though. No man-made lights skewing the white balance. TV says it can do 10 bits, but if that might cause issues, I’d rather see what the audience does whenever possible, so I’m assuming I want rec709 4:4:4 8 bit here, right? Again, different rec709 and related options available.

6) TV calibration. I’ve done a few calibrations before, and I have a colormunki, but it’s been a while. I’ve read that the tv has an “11 points white balance control”, gamma setting, and so on, so I should be able to dial it in with the colormunki. The set is known for being remarkably accurate within the standard gamut with the exceptions of very dark colors, and brightness inconsistencies from the edge lighting. Is there something I should be aware of here regarding target white balance, gamma, or anything else (again, in a room with indirect sunlight), or just follow the colormunki wizard?

Really appreciate the help, and hoping I can finally get this stuff straight after working for years in less than ideal circumstances. I understand this is still not a professional setup, but if I can get each of the points above worked out, I should be much better off than I have been.

[Greg Sage]

OK, thx. I vaguely remembered being told a while back that I needed to increase something in my “working color space” to avoid certain issues when colors combine or have effects added. I guess it was just the bit depth.

In practical terms: Even just in the defaults, there are several options for rec709, and is increasing the bit depth to 16 bits merely an internal affair inside AE, or does seeing it accurately as I’m working necessitate different card settings, increase HDMI bandwidth, and / or affect other items. The TV, for instance, is listed as being capable 10 bit color. Is that an issue?

[Greg Sage]

Interesting trick about stretching and back. Hadn’t heard of it. Might be just the ticket (in conjunction with frame averaging to simplify and denoise each frame). Just to be clear, you’re talking about making it half speed, then back to normal, right? Just using the built in AE tools to do so? No 3rd party plugs to dial in the process?

I’d think that would be better off BEFORE the effect since any jumps in threshold would have already triggered jumpy lines by the time the stretching is done. Point is to smooth it out so those jumps don’t occur if possible.

Also, I came across RE:Vision plug called DE:Noise Frame Average which seems to be what I was describing about taking out any chatter that’s not persistent across x frames. Any particular reason NeatVideo would be better?

[Greg Sage]

Any frame rate… and it’s not just one plug I’m talking about. Essentially any effect that redraws based upon threshold rules, so any of the cartoonize (inside AE or external) fx that draw cartoon lines around things, or convert the picture to sketch (running through PS), or any of the brush or paint dab fx, etc.

Essentially, I think I need some sort of frame averaging de-noising with a deep setting, but I don’t think that’s going far enough. Is anyone using something like this, but then with some particular combination of settings, or in combination with something else to smooth out any motion with minimal blur?

For instance, if you’re slowing something down below the actual frame rate, Twixtor interpolates the frames and at least tries to redraw the image at various interpolated stages to fill in the correct motion between the known frames.

Is there some sort of intelligent frame averaging that does something similar, but with actual footage that’s not being time stretched, so it looks at actual frames, yet shifts them around a bit to give best fitted curve to all motion based on surrounding frames?

If not, what’s the most advanced frame averaging effect that can denoise intelligently and give a nice clean output with minimal blur? That’s at least a start.

Surely anyone that’s ever used any sketch effect or anything like it has had the same thought, that it’s a bit chaotic and would be much nicer if the motion was slower and smoothed out… I’m just trying to figure out how to actually dial that in.

[Greg Sage]

Yeah, thx.

I did all that stuff, but no dice.

I see now that they reintroduced the filter in the sub-version just after the one I have installed. Real PITA right now trying to figure out how to update anything due to EEA managed package install that won’t allow me (even though I’m the admin) to do any updates.

[Greg Sage]

No such option under stylize on any of the 15 installs I have for cc2014 or 2015. Is there some sort of download pack you’ve installed?

AFAIK, this hasn’t existed as a stock item since CS6 or so.

EDIT…. Ahhhh… I see now. It hadn’t occurred to me that they would have actually caved to public pressure since then and RE-introduced it after officially cancelling the whole pixel bender aspect.

Apparently, I’m running the version of cc1015 just before it was re-introduced. Updating now….

[Greg Sage]

No such option under stylize on any of the 15 installs I have for cc2014 or 2015. Is there some sort of online filter pack you’ve installed?

AFAIK, this hasn’t existed as a stock item since CS6 or so.

[Greg Sage]

Yeah. Thx. That’s getting the audi marked, though… which is the easy part. It might save a little time on that bit, but the big question is how to essentially mark the footage that’s not to a beat and reallign those markers to match the marked audio.