Color Grading on Mac

MOV or other container should be really irrelevant as the same color metadata should be in actual codec (VP9/h264/h265/ProRes etc.) headers and through the browser's video engine passed to OS color management.

that's perhaps your private opinion, but in fact it can be defined in both ways, and the ]'colr' parameters atom in quicktime and its corresponding slightly modified counterpart im MPEG-4 are in fact the more common variants and in fact prioritized by the ISOBMFF standard (see: https://poynton.ca/notes/misc/sde-nclc-vui-nclx.html). it's also a more general kind of description, which can be adapted to different file formats without additional code, precedence issues and different ways to specify color information in every codec again (see: https://github.com/Netflix/vp9-dash/issues/5 for a real world conflict of this kind). i really understand your POV, but you will hardly change all the complexity and historically grown divergences in this field solely by your convincing voice.

but in practice you will hardly see any issues related to this topic in .mp4 files. it's more or less only apples crazy practice, to make anything as incompatible as possible -- like forcing a rec.601 interpretation instead of rec.709 for HD content, if not explicitly defined otherwise --, which makes the whole story very difficult and incompatible in practice.

It's all doable, but not in anyones interest...

many aspects of this issue are really hard to change or improve, without inducing new incompatibilities. it reminds me of this annoying off-by-one synchronization issues, which plagued the old QT implementation from apple for a long time. all other software had to fix it by ad hoc hacks, to provide compatibility. an now, this old QT implementation is slowly vanishing and the new apple mediaframwork doesn't show this issue anymore, but we suddenly face off-by-one issues in the other direction, because the old questionable 'fixes' are still present in most non-apple software. it's really horrible practice, to solve compatibility issues, which are in fact just caused by one proprietary software and its bugs, in such a manner. and i really don't understand, why ProRes, which is in fact the file format, which undoubtedly should be seen as the most affected one of this kind of issues, is still commonly seen as the gold standard of professional video file exchange?

that's perhaps your private opinion, but in fact it can be defined in both ways, and the ]'colr' parameters atom in quicktime and its corresponding slightly modified counterpart im MPEG-4 are in fact the more common variants (see: https://poynton.ca/notes/misc/sde-nclc-vui-nclx.html). it's also a more general kind of description, which can be adapted to different file formats without additional code, precedence issues and different ways to specify color information in every codec again (see: https://github.com/Netflix/vp9-dash/issues/5 for a real world conflict of this kind). i really understand your POV, but you will hardly change all the complexity and historically grown divergences in this field solely by your convincing voice.

I don't see any problem. Codec header should always be most important. Technology is there, just no one cares to use it properly.

but in practice you will hardly see any issues related to this topic in .mp4 files. it's more or less only apples crazy practice, to make anything as incompatible as possible -- like forcing a rec.601 interpretation instead of rec.709 for HD content, if not explicitly defined otherwise --, which makes the whole story very difficult and incompatible in practice.

Really? What Microsoft has done about it?
I actually can flag Rec.709 2.4 gamma based file on OSX which will be properly displayed by OSX default player on P3 screen?
Can you do it on Win10?

I don't see any problem. Codec header should always be most important.

sorry, but it's in fact exactly the other way around!

in http://standards.iso.org/ittf/PubliclyA ... 2_2015.zip on p.158 it clearly states:

"If colour information is supplied in both this box, and also in the video bitstream, this box takes
precedence, and over‐rides the information in the bitstream."

someone else grades a third clip and grades the same client's logo to match visualy on their iMac screen

Completely bogus argument. In the absence of any colour management, the logo will look identical as a png/bmp/tiff file on disk to whatever you're "grading" it to, as long as both are viewed on the same monitor, no matter how bad said monitor is. If it matches there, it will still match on a reference monitor.

Mischa i'd agree with you if it was as simple as matching a color swatch,

unfortuantly we usualy get a "style guide" that the agency's client prepared, usualy has CMYK values, pantone ref number, and samples for layout..

what we usualy do is create a perceptual match, look at the sample in daylight, office light etc, and tweak knobs until we have the agency art director approvals

the chance of getting approvals by mathmaticly matching the sample injested into a gradeing system is about zero, they will not be close

so diffrent colorists working in diffrent cities will be close if useing a callibrated mon,

doing a perceptual match without a solid base to work from is a fools errand, and likely will lead to the agency never entering your doors again...

i have this opnion as i'm the person who's doors they do walk through after they have been burned by someone gradeing on an iMac screen alone...

I don't see any problem. Codec header should always be most important.

sorry, but it's in fact exactly the other way around!

in http://standards.iso.org/ittf/PubliclyA ... 2_2015.zip on p.158 it clearly states:

"If colour information is supplied in both this box, and also in the video bitstream, this box takes
precedence, and over‐rides the information in the bitstream."

Ok, if they decided so. It doesn't change anything- technology is there.

someone else grades a third clip and grades the same client's logo to match visualy on their iMac screen

Completely bogus argument. In the absence of any colour management, the logo will look identical as a png/bmp/tiff file on disk to whatever you're "grading" it to, as long as both are viewed on the same monitor, no matter how bad said monitor is. If it matches there, it will still match on a reference monitor.

so diffrent colorists working in diffrent cities will be close if useing a callibrated mon,

doing a perceptual match without a solid base to work from is a fools errand, and likely will lead to the agency never entering your doors again...

i have this opnion as i'm the person who's doors they do walk through after they have been burned by someone gradeing on an iMac screen alone...

Yes, but we don't really argue that we don't need reference monitoring. Question is do you really always need it?
What we definitely don't need is Dolby monitor (which is btw now getting outdated)- looks like for many people answer for all monitoring problem is- you need Dolby monitor. Sorry, but this is such a nonsense.

I can give you opposite example.
Very expensive project graded on Dolby by highly respected colorist had to go for another correction because client did not liked how it looked online on Mac screens. Client did not care about Dolby look (not a single end user will see it this way) as 95% people will see on their laptops (and at least Mac screens have quite consistent look).
This is getting more and more popular problem for studios. And in this case we are talking about serious projects, which 95% people from this forum have nothing do with.

I'm almost sure that most problems don't come from fact that people use uncalibrated monitors, but from all other problem with consumer viewing technology.
If you can afford it get something fairly good, calibrate and have this reference ready to use.
If you can't afford it then you can still work around it. There will be most likely other issues on the way which may cause bigger problems than lack of perfect monitoring.

i'm in the crowd that reguaularly uses Dolby, X300, Christe and the king kong.. the Barco laser

along side PanaPlasma, Sony and Flanders OLED's...

but really a decklink card + a Flanders LCD = <3k

so that seems to be what a basic level of being able to trust your work will cost, guess the question is what the cost is of disapointing decent clients is, and what is a better solution...

but really a decklink card + a Flanders LCD = <3k

so that seems to be what a basic level of being able to trust your work will cost, guess the question is what the cost is of disapointing decent clients is, and what is a better solution...

I think the point in this case is that quite a few users of Resolve don't have "clients," instead they are hobbyists, independents, or posting videos on youtube to bring in some money from ads, for example.

I totally agree that if someone's paying you to produce videos, you really need to use a reference monitor. If you're collaborating with others, you need to use a reference monitor. And I'd argue that even if you're not being paid or not collaborating with anyone, but you're serious about the quality of your work, you are still better off getting a calibrated reference monitor -- if for no other reason than to have a consistent reference point across projects or within the same project if you're working on it for many months or years, and to reduce time in troubleshooting (since you can take the monitor out of the variables that might be causing the problem).

The question is more like "if you're working by yourself, don't have clients, and the final destination of all your work is youtube and/or Vimeo, do you really need a calibrated reference monitor?" The answer is probably "no you don't need it, but you'd still be better off with one if you can afford it." Here in Canada, a Flanders DM 240 is $5,500 Canadian plus duties, taxes, and shipping, which brings it over $6K. Even the smaller 16.5" DM model is $5K CAD all-in. I'm going to get one, although I have to admit I've been eyeing some of the cheaper alternatives such as HP's 10-bit monitors.

This is good summary.
You could only add- even bit worse monitor/TV but calibrated will be better than something "apparently" good, but uncalibrated/validated. If you do trust something at leas you should know that it can really be trusted (and how good it's). Same applies to the whole workflow.
I would definitely look at Eizo or NEC instead of HP.
Other valid option is good TV. For non-critical work this is definitely good choice which not only covers UHD, but also can cover HDR needs at fairly low price.

@ Brad, wow i'm shocked at the Flander's priceing, thought they were MUCH less costly... i have not been in the market to purchase one, i have freelanced in a room with one, and it was fine for TV..

i'm seeing the attraction of a LG|C8 + lut box now, that's where i'm likely going for my personal suite

@ Brad, wow i'm shocked at the Flander's priceing, thought they were MUCH less costly...

Well, keep in mind that those are Canadian dollars...in USD it's more reasonable. The exchange rate kills us up here, plus I'm in Québec where sales tax is about 15%. I do have my eyes on an Eizo CG247X, which is a bit more reasonable at around $3K CAD.

Although somewhat OT, I'm curious about other people's settings for projects requiring web deliverables only (when using a DeckLink card and calibrated monitor):

These are mine:

Project Settings > Video Monitoring > Data Levels: Video
Project Settings > Color Management > Timeline Color Space: Rec.709 Gamma 2.4
Deliverables > Advanced Settings: Data Levels: Full

I find it interesting that the 'YouTube' and 'Vimeo' Deliver preset have Data Levels set to Auto.

'm seeing the attraction of a LG|C8 + lut box now, that's where i'm likely going for my personal suite

I've been using Calman Studio in conjunction with an eeColor 3D Lut Box 65^3 for my large consumer client monitor. Inexpensive and works well -- but limited to HD resolution. Haven't yet found a fairly priced 4k/UHD box with a large cube and support for HDR, but honestly haven't looked for a while.

Although somewhat OT, I'm curious about other people's settings for projects requiring web deliverables only (when using a DeckLink card and calibrated monitor):

These are mine:

Project Settings > Video Monitoring > Data Levels: Video
Project Settings > Color Management > Timeline Color Space: Rec.709 Gamma 2.4
Deliverables > Advanced Settings: Data Levels: Full

I find it interesting that the 'YouTube' and 'Vimeo' Deliver preset have Data Levels set to Auto.

Don't use Full Levels with YUV based codecs. You are asking for problems. Stick to Auto, which is Video.

And many sound people have a cheap Ghettoblaster sitting next to their expensive speaker system just because of that…

Not exactly, but for many years, they did have full-range studio monitor loudspeakers, plus Auratone (aka "AwfulTone") speakers on the console just to give them an idea of "the average bad speaker." Starting in the late 1980s, some used the Yamaha NS-10M's for the same thing. But in that case, so much depends on room acoustics, sound can change radically just depending on how big the room is, how it's designed, and where you're sitting relative to the speakers.

And there are similar principles for video monitors. If the client is judging the work on a crappy display in a room with daylight streaming through their windows, good luck. Or if the display is over here, and the client is 10 feet to the left... it'll look very different than it does head-on.

Very expensive project graded on Dolby by highly respected colorist had to go for another correction because client did not liked how it looked online on Mac screens. Client did not care about Dolby look (not a single end user will see it this way) as 95% people will see on their laptops (and at least Mac screens have quite consistent look).

It's very hard to deal with massive ignorance from clients. One of my saddest phrases is, "there's nothing worse than a non-technical person in a position of power."

All you can do in this case would be to take a Mac and try to calibrate it as close as you can to something approaching Rec709. But somebody needs to explain to the client why technical standards exist and how the world really works. (There's a tale I could tell about a huge, major A-list director who demanded he approve his billion-dollar feature on 20 different monitors, but that's a long story.)

BTW, Macs do not have a consistent look. I tell this story often, but I'm at the Grove Apple Store in LA at least once every month or two, and every time I'm there I chuckle at the 6 iMacs in a row, generally playing the exact same demo, and they all look radically different from each other. If Apple can't get their own brand-new displays even close, in their own store, what makes you think they're going to look similar in the real world? This is why actual standards exist. And Apple (or their suppliers, Foxconn and Samsung) make zero effort to keep their displays accurate, beyond some fairly gross tolerances like maybe 20%. We need tolerances at least down to maybe 2%, which ain't gonna happen for a $200 display in an iMac.

Well- 3 Macs (I checked also my gf work one) at my home are all different years/models and they all match very well as for end consumer needs.
2% for home is unrealistic. This is probably even not achieved in broadcast/post houses or cinemas (if you would go and measure it all now).

Well- 3 Macs (I checked also my gf work one) at my home are all different years/models and they all match very well as for end consumer needs.

I suspect we have different ideas as to what matches and what does not. If it works for you, fantastic -- you have a great life.

Well- 3 Macs (I checked also my gf work one) at my home are all different years/models and they all match very well as for end consumer needs.

I suspect we have different ideas as to what matches and what does not. If it works for you, fantastic -- you have a great life.

It seems to be that Andrew is referring to perceptual matching, not the actual mathematical measurement.

It says clearly: "match very well as for end consumer needs", which is not trying to get deltas <3 etc.

It's not even about watching something on 2 devices next to each other, but watching on 1 of them and then on another. If you don't notice/remember difference then this is already good start. Unfortunately by average this is still not the case for most TVs or monitors.

Besides- does Dolby look like Sony BVM in terms of actual "what you see"? No, they are calibrated, yet so different. Which one is "correct" ?