Calling ALL Colorist...

[Marcos Montenegro]

Very interesting and informative post. I'm a one man show (director, writer, DP, actor, colorist), and I'm halfway through an independent project. So far, the clips I've edited and consequently shown to friends in the industry have gotten great reviews in terms of grading, but I have not had a chance to show them through a large enough projector to show possible errors, which I'm sure I've made.

Regarding grading on a budget, as some have stated:

1) What monitor would you recommend to those that are on a budget, say under 2K? I had this in mind:
http://www.bhphotovideo.com/c/produ..._SV_MultiSync_PA271W_BK_SV_27_Widescreen.html
2) What calibration software/hardware?
3) Any NTL (Premier, MC, VEGAS) internal tools that can be helpful in calibrating and grading properly?

Thanks!
Marcos

 

[Zakaree Sandberg]

get a really good graphics card, get an NEC monitor, calibrate with spectraview, nail your grades..

done deal

 

[Lou Cann]

Ok, everyone's post will be printed out. When I have some quiet time to myself (kids are driving me nuts) I will use these posts to help me better understand this wonderful world of filmmaking. I hope that I CAN REALLY count on HELP from this Forum in the future. I really want to Learn. Next week I will post some of my images & files. I am hoping that you guys can help me understand better.
-----------------------------
FINAL THOUGHTS:
1. What you said is SO true. My "inability" to grade DID EFFECT not just myself, but the filmmaker as well.

I was embarrassed when I saw the film at the festival. I had NO ANSWER when asked "If it's the Projector that's the problem, then why did everyones elses film look better than mine?" MY only response was..."take the dvd home & tell me if it looks bad"
All that did was add to his confusion. He started to think it was HIMSELF as a Filmamaker that might be the problem.

But in all fairness to myself, I "thought" I knew what I was doing. After all, I read about 6 books, 50 tutorials online (ie FXPHD stuff)
I have been grading for a year, so why wouldn't think I could handle a film festival?

I do NOT want & WILL NOT make that mistake again.

I can't afford to go to school to learn, so forums & the internet are my only source.

Thank you guys for all your help.......and yes..even you MIKE MOST.

 

[Martin Weiss]

I moved this thread off the RED discussion, as it has nothing to do with RED.

 

[Marc Wielage]

So I ask you....What in GODS name am I doing wrong?

Mike has pretty much told you the truth. Even if you don't like it, what he says is essentially right. It's a bad idea to try to use something like an HP 2408 display for critical color-correction. Don't get me wrong: it's a great computer monitor, and I use them myself, every day. But I'd never trust it for making color decisions.

Three suggestions:

1) hire somebody qualified (like an ISF-certified tech) to come in and calibrate a real monitor, to show you what actual Rec709 should look like. You can't use charts and eyeball the monitor to adjust it; you have to use scopes and measuring instruments to get it right. Getting accurate grayscale results is particularly tough.

2) book a few hours at a post house and sit with an experienced colorist, go through your footage, and set some experimental looks as a test. Compare what they do to what you have.

3) read Alexis Van Kurkman's book Color Correction Handbook: Professional Techniques for Video and Cinema. I don't think you can learn color correction from a book, any more than learning how to ride a bicycle from a book, but it will at least give you the basic concepts involved, show you good pictures from bad, how to read a scope, and all the basics. Hurkman does a terrific job, and even I learned a lot from what he had to say, despite working in post here in LA for 30 years.

 

[Gunleik Groven]

Thank you guys for all your help.......and yes..even you MIKE MOST.

You probably have no idea how fortunate you are chatting directly with Mike and getting good advices from him here...

His advices are qualified, and if you disagree, make sure you have a documentable way to do so...

I'd like to stress his last advice:
Do complete workflowtests for the media you are delivering.

If it is for the net, try it on the net and watch it on multiple screens and computers
If it is for the cinemas, take that little DCP to test it and compare
If it is for TV... You get the pattern?

Calibration is an important factor of a signalchain, but you still need to tune your eyes into what you can expect from a scene under different conditions.

 

[Subhadip Sen]

If you don't want to hire someone to do it, you have to learn to calibrate your display correctly. You can get a commercial colorimeter like X-Rite Display 2 for less than $150 and free software like HCFR to go with it. It is inexpensive and an essential cost. This is not optional. Looking at charts will get you absolutely nowhere. As for monitor, Asus PA246Q is an excellent option on a tight budget. Being a 10-bit monitor, you should use Quadro GPUs. Use a calibrated P-IPS panel monitor, follow the scopes (learn your scopes thoroughly - very important), and you will see major improvements in your work.

Of course, that's not where it end. You have to know your display types (how the display renders colour), colour spaces, colour theory, colour as detected by your eyes etc. I am not even going into the craft of grading here - just the technical bits. The problem beyond all of that is that for every calibrated display, there's a million miscalibrated ones. By its very nature, TN panels (present in almost every laptop and most desktop screens) have narrow colour gamuts, high black depths and 6-bit colour depth. Unfortunately, there's no easy answer as to how you can compensate for this, and as far as I am aware, there's not much documentation on the variance in displays. The only suggestion I can offer is extensive testing, learn scope patterns and variances for the same - as Gunleik has suggested above.

 

[Jake Bastian]

But in all fairness to myself, I "thought" I knew what I was doing. After all, I read about 6 books, 50 tutorials online (ie FXPHD stuff)
I have been grading for a year, so why wouldn't think I could handle a film festival?

I think this is where a lot of us (including myself) go wrong. Book and tutorials are good, but they don't make you a pro. I think the best advice is something Mike said to me when I was asking about books on color matrixing. Find some way to hire a pro to work on your stuff or a consultant to come in and see how you work and then watch what they do. That kind of experience isn't something that can be replicated by watching a tut or reading a book.

 

[Ethan Cooper]

All the big guns here keep stressing the importance of scopes but are you guys talking about hardware scopes or software scopes? If software, what are you guys using? (OSX here)

 

[Iannis Holwech]

As a "hobbyist" on a budget, buying a calibrator (before buying a new display), like the new positive reviewed X-Rite i1Display Pro ($249 at B&H) would be the first thing to do.

Then you can check the difference in your graded material between the old non-calibrated display and the calibrated. Should be a great learning experience.

And you will find out if your current display can be calibrated accurately.

 

[Scott Crawley]

Without knowing precisely how you went about set up and testing, folks here can only guess in a general sense as to where you screwed up. You need to provide more specific information about your methodology and equipment. What charts did you use exactly and where did you get them? How did you display those charts? What was the signal path? What monitor profile are you using? Am I to understand that you downloaded some jpegs from the Internet and opened them directly in Resolve?

You have apparently set your monitor to be overly contrasty somehow, but that is the only conclusion to reach given what you have told us.

As some of the others have mentioned, for as little as $150 you can get a colorimeter and software to better profile your monitor. It's a must. Think of how much time and money the film makers invested in their project, and the look of their production all ends with you. Now are you willing to spend $150 on a basic tool?

 

[Hrvoje Simic]

Asus PA246Q is an excellent option on a tight budget. Being a 10-bit monitor

8-bit + FRC.

 

[Subhadip Sen]

8-bit + FRC.

Despite the method used, the monitor does simulate 1.07B colours. A-FRC in P-IPS panels is very effective and even to the most sensitive of human eyes, the dithering is completely invisible. Using a 8-bit GPU outputting 16.7M colours will leave it at a disadvantage.

 

[M Most]

All the big guns here keep stressing the importance of scopes but are you guys talking about hardware scopes or software scopes? If software, what are you guys using? (OSX here)

Scopes basically provide two things: an absolute indication, without the possibility of subjective interpretation or calibration issues, of what your signal parameters are, and a sanity check for what your picture monitoring is showing you. Your eyes can be fooled, your scopes can't. And your monitor can be misleading, the scopes can't. Assuming your scopes are being fed the same signal your monitor is getting, they are always right. If they show that you have video levels of 20 units on objects that are supposed to be bright white and your monitor is showing a very contrasty image, your monitor is wrong. Conversely, if your monitor is showing you an image with visually rich blacks and your scopes show your blacks at 25 units, your monitor is wrong. That's what I mean by sanity check. Scopes are, of course, also very useful in checking things that your eyes can fool you with, such as critical black balance when it's needed.

Hardware scopes are valued by colorists because their displays are usually cleaner, finer, and more precise. The display can usually be expanded and in that mode can show very critical values and details. Software scopes, in general, have traces that are less fine, so it's more difficult to determine exactly where things like low level black details are. This is usually not that big a deal, but with really dark images it can mean the difference between perfectly balanced blacks (and thus a more pleasing image with more depth and better separation) and slightly misbalanced ones. Some software based units, like the Omnitek products, strike a good middle ground, giving the convenience of running on commodity hardware and standard monitors but with much finer displays that give you better feedback. The Blackmagic Ultra Scope package is pretty good and very affordable, but it does suffer a bit from some of the things I just mentioned (lack of vertical expansion on the parade display, lack of fine details in the displays). Same with the ScopeBox package. The rasterizers from Tektronix are pretty good, but Omnitek hits a real sweet spot and has gained a lot of traction in major post facilities for that reason (along with optional 444 capabilities).

 

[Lou Cann]

This is "ONE" of the charts I used http://baranik.com/blog/wp-content/uploads/2011/03/test-chart2.gif
to calibrate my monitor.

When I view my projects on MOST tvs, it look pretty close to my monitor. The HUGE ISSUE is when it goes to the theater.
That's when all goes to Shit..

I have to admit, I am very weak in knowledge when it comes to color space & Color management. Again, I ASSUMED that XHA1 fooatge & 5D footage doesn't require much color management as Film scans.

Again, I don't know much.

i HAVE SOME QUESTIONS:
-------------------------------------
When 5D fooage come into a program, I assume how it is displayed is correct (for 5D). And that I didn't need to switch color spaces (ie Log to LIN )

Am I Correct?
Do I have to change it from default (Linear ) to Rec 709?

 

[M Most]

This is "ONE" of the charts I used http://baranik.com/blog/wp-content/uploads/2011/03/test-chart2.gif
to calibrate my monitor.

When I view my projects on MOST tvs, it look pretty close to my monitor. The HUGE ISSUE is when it goes to the theater.
That's when all goes to Shit..

Using charts developed for video display on a computer monitor will not work. For one thing, the gamma is different (even if you're set up for 2.2 or 2.4 gamma on the computer) and for another, the saturation difference is considerable and not linear across the color spectrum (i.e., reds are much more saturated on true Rec709 displays than they are on computer displays). You cannot set any of this up by eye and expect to be right. What's needed is a measuring device that isn't the human eye, one that does a subjective evaluation of the display across the color spectrum and across the luminance range. That's what a probe, along with calibration software, does. It uses the expected values of the standard you want to emulate and compares that will the values it measures on your display, across a wide range of known values. It then creates either a profile or a lookup table to force your monitor to look like the known standard. There is a lot more involved in this than simply setting black and white chips, especially when the gamma is different. Your method would actually be more accurate if you were using it on a standard television monitor, because that's what it's designed to be looked at on. A television is designed, by definition, to display images reasonably close to the Rec709 standard, unlike computer monitors. So if you can't calibrate a computer monitor properly, and/or don't have one that can properly emulate the color gamut, gamma, and greyscale of a Rec709 compliant display, you're better off using a properly set up television for monitoring your images, assuming it's set up sensibly and you're accounting for proper SMPTE scaling in your output path. Digital cinema is another story. It uses a wider color gamut, but more importantly and significantly, it uses a gamma of 2.6, a lower brightness level than video monitors (about 14 ft/L compared to about 30 ft/L) and is designed to be seen is a darkened room. Emulation of those conditions is possible using calibrated displays, but the only accurate way of viewing digital cinema is to view it under theater conditions. That's why projection based digital intermediate theaters exist in facilities dedicated to that kind of work.

When 5D fooage come into a program, I assume how it is displayed is correct (for 5D). And that I didn't need to switch color spaces (ie Log to LIN )

Am I Correct?
Do I have to change it from default (Linear ) to Rec 709?

It is not Linear. Like many people, you are regarding anything that isn't Log as Linear. That is not the case. Video images are not Linear, they are gamma encoded. In other words, they have a gamma curve applied so that they look correct on video monitors. True linear light images will not look correct on a video monitor because they do not have a gamma curve. You can see this by using the Linear setting in Redcine X. That yields a true linear light image, which on any common monitor will look much too dark.

Back to your question. What you see is what you get. The Canon cameras, when set up with their normal settings, produce a video gamma encoded image. They should look correct on video displays without any further manipulation, other than perhaps black and white adjustments (they're often a bit extreme due to their use of full range rather than SMPTE scaled values). The Technicolor Cinestyle setup files emulate a log space for those cameras, so if you use those, you need to use their corresponding Cinestyle to Video LUT in post.

 

[Lou Cann]

Hi Mike Most,
Ok I am going to do A LOT of research on this issue.
In the meantime, I need some sort of a starting point for myself.
-------------------------------------------------
For example,
The guys over at New filmmakers of New York are pretty cool over there.
I could go there before they open and test my footage.
They will help me with any questions or problems.

I know that my grading appears very dark, ( flat, low contrasty) on their screen
If I make up a LUT that simulates the theater's screen, and grade accordingly, will that be a step in the right direction?
I would then bring over some test shots to see if it looks better.
Is this acceptable or is this just stupid?

Again, TV isn't my issue. It's THEATER.

I will purchase some of the equiptment you said to use, however, I still need deal with THEATER.

Thank You

 

[Matt W.]

I asked elsewhere, but Mike, how would you recommend calibrating a consumer HDTV to rec709 (or whatever is appropriate) for a grade (in Apple Color) which will then be output for broadcast and also 2k DCP? No chance to see the screen before the film is projected so this thread has me worried about gamma issues I might encounter along the way. Thanks...

Mostly just interested in following this thread.

 

[Lou Cann]

.

For most projects you should make sure all your black details are above 0 IRE and all your highlight detail is under 100 IRE on the parade scopes.

.

Ok, Although I get that, I still am confused.
For Example, If there was an image of a room with a mirror in it and 90 % of the RGB values were 160, 161, 155 (8 bit)
But there was one little reflection on the mirror that had values 255,255,255. So the overall Parade would show a spike in that one area.

My reaction would be, to raise the luma values of 160, 161,155 to a higher value, say 220, 221, 205 Now that would mean the mirror reflection would have clipped , but the overall scene would be in legal values & overall a brighter image.

Is this acceptable to do that?

When I read a scope (Parade), I check to see just "What" areas are clipping. If the parade gets red lined because of a tiny area that doesn't effect the overall image, I ignore it. (as long as it's tiny or doesn't effect the look)

Again, do you guys do that too?

 

[M Most]

Ok, Although I get that, I still am confused.
For Example, If there was an image of a room with a mirror in it and 90 % of the RGB values were 160, 161, 155 (8 bit)
But there was one little reflection on the mirror that had values 255,255,255. So the overall Parade would show a spike in that one area.

My reaction would be, to raise the luma values of 160, 161,155 to a higher value, say 220, 221, 205 Now that would mean the mirror reflection would have clipped , but the overall scene would be in legal values & overall a brighter image.

Is this acceptable to do that?

When I read a scope (Parade), I check to see just "What" areas are clipping. If the parade gets red lined because of a tiny area that doesn't effect the overall image, I ignore it. (as long as it's tiny or doesn't effect the look)

Again, do you guys do that too?

You seem to be getting hung up on numbers. The numbers are only important when you're exceeding their limits. What's important is the image, which is why a properly set up monitor is so critical. Some images might look great with a maximum luminance value of only about 50 units. Other images might look very flat under those conditions. It all depends on the content, the photography, the overall image contrast, and the mood you're trying to set.