I apologize if this question has been answered elsewhere, but I have searched and not found much.

I have two monitors, a Dell U2410 (wide gamut), and an older Dell TN-panel (will probably be replaced by a U2711). I want the best possible setup for accurate color reproduction when editing video (and ideally photos as well). It seems it should be possible to calibrate both displays and get accurate video color reproduction using only the monitor profiles, considering this is clearly possible with still images (although much trickier in Windows vs. OSX). However, advice I find is all over the map.

Is the recommendation to use an external monitor strictly because Premiere and other NLEs are not color managed? FCPX suggests that on a well calibrated monitor, colors will be accurate within the NLE and a dedicated monitor isn't necessary. This would be the ideal configuration for me, although I don't really want to use FCPX.

The U2410 covers a lot of gamut, so it seems like it should be able to get pretty close to any target after calibration, even if there is some dithering or banding.

Any enlightenment on this topic would be appreciated!

Here's a very concise guide to using color bars; it's an Apple document but it applies to monitors in general: Color User Manual (http://documentation.apple.com/en/color/usermanual/index.html#chapter=A%26section=2%26tasks=true)

and another: http://spareroommedia.com/video/monitor_setup.html

There are several other similar discussions, Google "calibrate video monitor color bars" and you'll see several. But these tell you what you need to know. HTH...

Will this is a tricky subject. In the best case scenario, you would have a hardware card driven external monitor for video as well as computer screens for everything else.

At the heart of the matter is the difference between video and computer signals. Video for television viewing is in the YUV color space. This means DVDs & Blurays should be created from a YUV space environment.

Computers use the RGB color space. So anything you view on a computer driven screen (including video) is being derrived from the RGB color space. RGB & YUV are different.

What to do? The fact you know they are different is the first step. I think Premiere might have some sort of YUV emulation in the preview monitor but I do not know for sure.

Do you produce for television viewing or internet or both?

Battle - Thanks for the links.

Tim - Mostly DVD/Bluray, but both. I will use an external monitor if it's absolutely necessary, but it seems like it shouldn't be -- a vestige of less advanced software having to punt to special purpose hardware. What I don't understand is why, if the monitor, OS, and application all know the gamut and color profile of both the monitor and the video signal, it cannot simply do the math and display the video signal as it's intended - RBG/YUV conversion is fairly simple. My understanding is that this is what FCPX does, although I don't use FCPX.

I prefer Premiere, but I am not opposed to working in other NLEs if they are more capable in this respect. However, work for some clients will require me to use other NLE software, part of the reason I want to make sure what I am seeing is accurate before sending it off.

but it seems like it shouldn't be -- a vestige of less advanced software having to punt to special purpose hardware. What I don't understand is why, if the monitor, OS, and application all know the gamut and color profile of both the monitor and the video signal, it cannot simply do the math and display the video signal as it's intended - RBG/YUV conversion is fairly simple.

A quick analogy using your argument above...

"I would like to print glossy stickers.

But I want a proof on flat white letter paper, not glossy heavy weight sticker stock.

Why can't you give me an accurate representation of how the colours will look and how shiny it will be?"

YUV and RGB have different colour space/reproduction. YUV (video) displayed on a native RGB device through an RGB connection (VGA, DVI, HDMI to a computer monitor from a computer) cannot ACCURATELY be displayed, IMHO. A wide gamut display receiving a YUV signal through a YUV connection (analog component, HD-SDI, HDMI at VIDEO resolutions from a YUV source) can be much more accurate.

I think Shaun has given a great example.

It is not about advanced or simple software it is about sending the proper signal to a proper monitor. Computers deal in RGB unless a specific device converts and distributes. Without that "simple" hardware card you will have a tough time getting an YUV signal from a computer.

Tim and Shaun - I don't think the analogy is accurate, unless you are talking about doing no YUV to RGB conversion at all. CMYK to RGB is a very common color space conversion used in print workflows, and calibrated RGB monitors can accurately match on screen RGB to a CMYK printer. YUV color space is no different. (Converting between Y'UV and RGB (http://en.wikipedia.org/wiki/YUV#Converting_between_Y.27UV_and_RGB))

And the conversion doesn't have to be done in hardware. All DVD/Bluray player software and most media players convert YUV to RGB, and these are used in calibrated theater environments. DI codecs are a mix of YUV- and RGB-based, so some conversion must be happening in NLEs that support them.

I understand that having an external monitor can be a matter of personal preference. But if it is not preferred, I don't see why color reproduction can't be just as accurate within the NLE. FCPX evidently manages it - is it a limitation of Premiere (or Avid)?

All DVD/Bluray player software and most media players convert YUV to RGB, and these are used in calibrated theater environments

What outputs are RGB on a DVD/BluRay player?

Analog component is YUV (despite having red, green and blue cable identifiers). HDMI can pass either but USUALLY passes YUV at TV resolutions.

I don't see why color reproduction can't be just as accurate within the NLE.

The output of your computer to its attached monitor is RGB. Therefore anything occurring on the computer screen is being output via RGB. A second output (for example, FCP's Digital Cinema output Full Screen) could potentially do a software conversion to bring it closer but if you are talking about the viewer window in the interface being reference quality as YUV, you are deluding yourself.

Close enough? Sure. Especially for web output but don't claim it is referenced, because it isn't.

Shaun - DVD/Bluray player software like Apple DVD Player for Mac, or PowerDVD, Arcsoft TMT, etc. for the PC. All of these apps do the conversion in software to accurately display YUV content in RGB space.

My understanding was that even if an HDMI monitor receives a YUV signal, most will still convert to RGB before displaying anyway. This may vary by device, however, and do I have experience with the higher end monitors. Either way, you can connect a Mac or PC to a calibrated HDTV or monitor via HDMI or DVI, and the RGB signal will be virtually indistinguishable from an unadulterated TUV signal.

if you are talking about the viewer window in the interface being reference quality as YUV, you are deluding yourself.

Why am I deluding myself? There is no technical reason this can't be done, and FCPX claims to do it. If you have an explanation, I'm all ears.

Either way, you can connect a Mac or PC to a calibrated HDTV or monitor via HDMI or DVI, and the RGB signal will be virtually indistinguishable from an unadulterated TUV signal.

Your experience is MASSIVELY different than mine.

Despite knowledgeable people suggesting the contrary, you appear to want to believe what you believe. Power to you. I shall not waste any more virtual "breath" here. The fact that IO devices such as the Matrox MXO2 series, the AJA IO series and numerous BlackMagic Design and others products exist to do EXACTLY what you are claiming can be done natively in software either means:
- these manufacturers are snake oil salesmen;
- I and MANY of my colleagues are gullible;
- Myself and Tim are correct; or
- you are correct and I and many others have been bilked of our money and have been made fools of, despite seeing results that are different from yours.

Shaun - I don't think I have implied that you or anyone is gullible or is selling snake oil products. If I have, my apologies.

There are many reasons to use an external monitor, so I don't understand why the suggestion that it may be unnecessary for some uses with the right software warrants such exasperation. Apple suggests the same with their claim that ColorSync can render reference quality within FCPX. I don't know where Premiere and Avid stand. Apple may be overselling FCPX, I don't know, but I haven't seen anything definitive on these topics.

The existence of a hardware solution doesn't mean software cannot be just as capable, as evidenced by many media players/encoders/streamers that create reference quality output that you are welcome to compare yourself. Whether accurate color-rendering capabilities are built into the latest NLE software, or if it is OS-dependent, etc, I don't know. This is why I started the thread. It's not personal.

What I don't understand is why, if the monitor, OS, and application all know the gamut and color profile of both the monitor and the video signal, it cannot simply do the math and display the video signal as it's intended - RBG/YUV conversion is fairly simple.

YUV is not a color space, but an encoding standard based on sampling an analog video signal.

RGB is not a color space, but a color model.

If you're mainly into high definition video, HDTV, e.g., you are looking at a color space commonly called Rec. 709. This color space is almost equal (for all practical purposes) to sRGB.

Theoretically, Rec. 709 can hold infinite colors, though there are many colors that the human eye can see that fall outside its domain.

Now, to answer your question:

An OS has to cater to many applications and processes running simultaneously, any of which might suddenly need to display a message or GUI or whatever. Not all applications are made using the same 'principles', and unless an OS has access to the source code it cannot know for sure what math is actually being used.

E.g., when you apply a plug-in to a video - the app that runs the plug-in just executes the code, but the app does not have access to the math. The OS does not have the code to the app or the plug-in.

The app has to create a 'mathematical environment' so that the plug-in can work its magic without any bottlenecks in the math. 'Good' applications today must adhere to two basic math principles - running on 64-bit OS and providing a 32-bit linear float environment. This ensures a level playing ground for the latest technological advances and algorithms. Two apps that adhere to this are After Effects and Nuke.

Mathematics is just a tool - where you reach will depend on how you start - floating point or not? Log or linear? Which algorithm to use?

It's like a dysfunctional family, with the OS as the father, the app as the mother who sleeps around with other OS's and the plug-in as the teenage kid on dope. They're all trying to get along. In comes the unknown guest video who can screw up or maintain this balance. The best everyone can do is to stay out of each other's way.

This can be corrected if each and every manufacturer in the pipeline were to open its hardware and software for the world to see - just like a functional family who chats openly over dinner and solves their issues.

This led to three major developments:

1. Apple - completely closed system - the Corleones
2. Windows - Semi-closed system with the OS/father taking charge - King Lear
3. Open source - the Simpsons

Some interesting reading here to help explain:

Color Models (http://software.intel.com/sites/products/documentation/hpc/ipp/ippi/ippi_ch6/ch6_color_models.html)

Sareesh: my statement of colour space was to indicate the reproducible colours within the colour encoding scheme (yuv versus rgb). Gamut would have been more accurate.

To get past the thought that software can make a display "accurate", realize that a monitor emitting light is a PHYSICAL property which is dictated by the Laws of Physics. The liquid crystals or diodes can emit light only so accurately and the backlight essentially controls how accurate the display can be. The quality of the backlight explains why the same IPS panel being used in $400 Dell/HP LCDs and my $2500 Eizo have such large differences in accuracy.

The sad thing about all of this is the iPad 3 for $400+ has a more accurate screen than most $2000 monitors AND its higher resolution.

FYI, I use Premiere and have my Eizo connected via DP to a Quadro FX3800, which allows me to display YUV or RGB natively as well as 10bit (but this is on Win7 and Premiere is not 10bit capable on Mac).

Search for the article on provideocoalition about Premiere CS5, DisplayPort and RGB/YUV - its also linked several times by other members here. It will explain why you would need an I/O card/box and 3rd monitor versus why you can do what I do. However, I do have a BMD Decklink Extreme 3D and a Panasonic 1710w.

Steve, thanks for the PVC article. Here is the link for anyone that doesn't want to search:

ProVideo Coalition.com: TecnoTur by Allan Tépper (http://provideocoalition.com/index.php/atepper/story/does_premiere_cs5_achieve_the_impossible_dream_for_critical_evaluation_moni/)

Just to clarify - I did not mean to imply that I think software can make any display accurate, only that software calibration can be just as accurate as hardware. Cheap-O Monitor X will look no better connected to a calibrated IO box than through properly color managed software.

What I gather is that FCPX is currently the only fully color-managed NLE, so it's the only NLE capable of rendering its preview output corrected for YUV *and* adjusted for a display profile. However, it appears that OSX's GDI is limited to 8-bit, so that could be a major drawback for some. It's hard to find information on this.

Premiere CS5/CS6, as I understand it, will preview accurate YUV/RGB conversions in 10-bit (in Windows with a 10-bit graphics card), but does *not* adjust for a display profile, and instead just assumes sRGB. So it will be as accurate as your display handles sRGB natively, but if you create a display profile using a colorimeter, those adjustments won't be accounted for in Premiere. However, it appears that SpeedGrade has the ability to import a color profile and use that as a LUT for grading. It seems like this could get you closer to what ColorSync does for FCPX, but I haven't found much else about it.

If I got any of this wrong, please let me know. I'm just trying to understand the cost/benefit to see where I should spend my (limited) resources.

You are mixing up ICC profiles normally used for printing and calibration for video which is a 1D or 3D LUT.

When calibrating a monitor in software-only mode, the calibration is applied through the graphics card and its always 'On' unless you tell the software to disable it or change to another calibration. In 'Hardware' calibration, the monitor stores and applies the calibration settings so you can connect the monitor to anything and the calibration settings can be used independent of the source.

Also, Premiere would never use sRGB because Rec 709 is the default color space used in most programs & devices (HD video is Rec 709 whereas Rec 601 aka SMPTE-C is SD broadcast).

Btw, calibrating to Bars is an old way that has been used to get a 'close enough' setting where colorimeters cannot be used (too expensive and/or most monitors don't support hardware calibration).

From what I recall during the first several months after release, a major drawback of FCP X was the inability to use a properly calibrated reference monitor.

Generally speaking, the use of display profiles in video editing, vfx, 3D and animation is to have each person's monitor within the pipeline appearing as similar as possible to one another, eg. the compositor applies a LUT so he can see how the end product will look .

...only that software calibration can be just as accurate as hardware.

Software, even prosumer ones, are already light years ahead of hardware. As a closet programmer, I always have to get software to 'dumb down' to accommodate hardware.

Practical colorimetry is based on human feedback, like for example CIE 1931/RGB. Software needs a feedback loop, it can't 'see' for itself.

E.g., if I build a new display and connect it to software, FCPX, Premiere, whatever - how will the software know what the display is outputting? The display's feedback data might say Red, but the monitor might actually be showing something else. This is why one needs a calibration tool. Even with the tool, the unique features of my monitor's gamut will render the software useless, unless -

As in my previous answer - it can 'keep an open mind' and 'get a bigger math boat' - which 32-bit does well enough, with 'float' being a sufficiently acceptable mathematical compromise.

Think of an NLE or grading tool as a sheep pen. The best it can do is to be as wide as possible and have the highest walls - the sheep are the algorithms that can go beserk sometimes - the walls (color space/model/gamut) keep them in, the size of the grounds (the math base) keeps them happy and healthy.

The weak link in the chain is the human. No two humans will see the same color, even under the exact same conditions. To make matters worse, an individual can look at one color at 2 pm and again at 5 pm and they will look different. The same is true of a display - it's a physical system whose properties change due to thousands of reasons.

At best, it's a fuzzy relationship. Software and hardware need human feedback to be accurate. Humans think hardware and software will do the job, and he/she can keep their eyes closed.

I would sincerely warn anyone of using their monitor as the benchmark or standard, even if the monitor is a DI grade P3 projector. My experiences indicate (at least to me) that following simple color space standards and judging with your eyes is the right way to go, and that monitors just have to get with the program.

When calibrating a monitor in software-only mode, the calibration is applied through the graphics card and its always 'On' unless you tell the software to disable it or change to another calibration.

Right, calibration is always on, but non-color managed apps typically assume the monitor is targeting sRGB, regardless of the actual target. A color managed app will know this and compensate to generate the correct output, while a non-color managed app will map to the wrong colors if your display profile doesn't target sRGB.

Also, Premiere would never use sRGB because Rec 709 is the default color space used in most programs & devices (HD video is Rec 709 whereas Rec 601 aka SMPTE-C is SD broadcast).

My understanding is that Premiere does not target sRGB, but it assumes your monitor is targeting sRGB, not Rec 709, and corrects from there. If you target Rec 709 during calibration, then Premiere will be overcorrecting the image because it is not color managed and has no knowledge of your display profile.

Right now I don't have the funds to purchase the necessary hardware to get the 709/yuv preview from my PC to an external monitor. I've found my best option under the circumstances is to trust what the scopes displays (waveform, vector scope, histogram) in the color correction module of the NLE are telling me, not the colors my eyes are seeing.

I also sometimes burn a disk to play on the HDTV in the living room. If you don't want to get (or can't afford it right now) the right hardware to display to a proper monitor, then trust the scopes, not your eyes.

And of course the gamma shift in Quicktime that occurs because of the two color spaces is a real pain, having to encode twice for 709 viewing on tv and RGB for the web. But you have to do what's necessary to make the footage look right.

I want the best possible setup for accurate color reproduction when editing video (and ideally photos as well).

There are only a couple of options if you want an all-in-one solution. Eizo makes a few monitors that can display in both Rec.709 (HD video) and other color spaces such as AdobeRGB (stills). HP used to make a monitor that would do that as well, I think it was a DreamColor, but IDK if it's still on the market. Neither the HP nor the Eizos give you any real-time tools like waveform monitors or vectorscopes IIRC. So all you get is the image itself.

Other than that, if you want WYSIWYG video editing (and you can't do accurate color grading without it, or at least I can't), you need a monitor that can display Rec.709. There are tons of these, search for "production monitors". Or, HDTVs of course, but if you go the TV route, be sure to get one that can be accurately calibrated (some have ISFccc calibration mode which makes it easier and more accurate).

There are a number of reasons to use a good production monitor. One is that it's big enough that you can actually see the video so you can do a better job of evaluating it. A small image that takes up maybe 1/4 or 1/3 of your screen (what many NLEs will give without constant playing with the interface) is a lot harder to judge than an image that takes up all the real estate on a 24" monitor. And a good production monitor will give you real-time tools like waveform monitors, RGB parades, vector scopes, etc. Most NLEs provide these tools, but they don't run while the image runs; the tools only update when you stop the image. While useful, not nearly so useful as real-time tools.

Sorry for resurrecting and old thread. I've got the same open questions for color accuracy on Premiere CS6/FCPX/Avid, and Apple Color/Speedgrade without an I/O card.

I know the best way it's to get a proper I/O that convert computer signal to video signal and send it to the monitor, but in my case I have an imac with only a displayport (no thunderbolt), so the only way of attaching a I/O card is by using an old MXO and it's unsupported on FCPX, Premiere and new NLEs.

I'm thinking of buying an HP Dreamcolor (that can work in HD 709 color space) and connect it directly with a displayport cable. Like most people on this thread I've read the provideo coalition post, and if CS6 can really do an accurate YUV to RGB conversion, then I don't see the need for an I/O (for me). I just want an accurate broadcast preview. Does anyone know if this it's true?

Even a tech from the HP department said that there's no need for an I/O card using the dreamcolor attached via DP or HDMI, if you're sending a true RGB signal.

FCP doesn't do a proper conversion for the full screen preview, FCPX I don't know (I don't use it), Apple Color afaik it's not capable without an I/O card, same thing for Resolve, but I wonder if Avid does it right, and Speedgrade or AE (as part of CS6) will do it right.

FCP doesn't do a proper conversion for the full screen preview, FCPX I don't know (I don't use it), Apple Color afaik it's not capable without an I/O card, same thing for Resolve...

Interesting Ivan, can you point to a resource where this has been proved?

So here's something, I was talking to a friend of mine who's a professional colorist, and he said the Ipad (that includes version 1), out of the box (i.e. no tweaking with apps, if that's even possible) is actually 99% REC 709 calibrated, and finds it true and accurate to his much more expensive monitors.

I'm likely getting an Ipad mini soon, and was wondering if anyone knows if, aside from the non-retina display and lower resolution (mini is 1024x768), they are basically the same as far as gamma/colors/etc.?

Interesting Ivan, can you point to a resource where this has been proved?

I get this information reading other forums & tech blogs, so maybe it's right, maybe not, I don't really know for sure. I wish it had been proved, and have reliable information so we don't have to speculate :)

- FCP: Anybody seems to agree that Digital Cinema Desktop preview without an I/O Card it's not accurate for CC. Even Apple said it on 2005. (Pro Video Coalition: TecnoTur by Allan Tépper (http://tinyurl.com/co8uvh9))

Although Apple's colorspace conversions are certainly proper and trustworthy when rendered via software, Apple's 2005 warning indicates that this is not the case with the realtime conversion done with the Digital Cinema Desktop feature. To my knowledge, this has not changed with any newer version of FCP, and if it has, no one from Apple seems to be flaunting it.

- Apple Color: Apparently the preview window can't be trusted as accurate. And you need a I/O card to send a trusted signal to the monitor. (Pro Video Coalition: TecnoTur by Allan Tépper (http://tinyurl.com/co8uvh9))

Although Apple's Color program handles all material exclusively in RGB, to my knowledge, the program still does not pipe its program output specifically or accurately to a graphic card output, but only to one of the professional i/o devices.

Also here:
Color Grading is done on a proper monitor | Biscardi Creative Blog (http://www.biscardicreative.com/blog/2010/05/color-grading-is-done-on-a-proper-monitor/)

- Resolve: I don't know, I don't use it. I remember reading on reduser forums that the GPU card can't be trusted, but I don't have the source.

- FCPX: Don't know. I don't use it.

- Premiere CS6: Apparently the YUV to RGB Conversion is accurate:
Adobe Community: How does P Pro handle YUV (YCbCr) and RGB color spaces? (http://forums.adobe.com/thread/825920)

But doesn't AJA have a $200 i/o box, if you have a thunderbolt Mac? So not really a huge additional investment at this point, is it?

But doesn't AJA have a i/o box, if you have a thunderbolt Mac? So not really a huge additional investment at this point, is it?

No, I have a mini displayport Imac (late 2009), no thunderbolt. I wish I had one, because there's plenty of low-priced options for this interface :)

So here's something, I was talking to a friend of mine who's a professional colorist, and he said the Ipad (that includes version 1), out of the box (i.e. no tweaking with apps, if that's even possible) is actually 99% REC 709 calibrated, and finds it true and accurate to his much more expensive monitors.

I don't have an ipad but that's very interesting. How do you connect it to get the preview? Or do you play the final output from a file directly in the ipad?

Yeah, bummer 'bout the mac. One of several reasons I just bought a mid 2011 :-). I think your options are limited to the $1000 Matrox something or other? Unless someone finally creates a backwards compatible thunderbolt adatper.

Re: Ipad. . .unfortunately not a perfect solution. Basically, I think you have to export a movie (this colorist friend recommends uncompressed 8 bit or 10 bit, depending on your original format, rather than Prores or anything else, to prevent color/gamma shift) and load it to the ipad and play it off there.

He also mentioned a pretty neat method where you take a color chart, import into your NLE/CC software, throw it up on your scopes, and apply a CC filter to it and tweak 'til the scopes say the chart is bang on (i.e. red is red, black is black, white is white, etc.) Then, you look at your monitor. Is your white a little on the greenish side (etc. etc.)? That's how inaccurate your monitor is. Now you eyeball and tweak your monitor to get the color chart to look right (if the whites are little green make 'em white, etc.). Obviously not 100%, not a scientific method, but better than flying blind and will probably get all of us who aren't doing high end commercial/broadcast/theatrical release work, "close enough".

Ok I just realized what I wrote up there and now I feel like moron.

So the Ipad obviously won't play 10/8 bit uncompressed, so you'd have to save your test files in a format it can work with, and deal with any resulting gamma shift. Like I said, definitely a poor man's solution.

Ok I just realized what I wrote up there and now I feel like moron.

So the Ipad obviously won't play 10/8 bit uncompressed, so you'd have to save your test files in a format it can work with, and deal with any resulting gamma shift. Like I said, definitely a poor man's solution.

Yes, not ideal but it should work from what I do (not high end broadcast stuff).

My only solution it's to get an original MXO (that's what i'll do + the HP Dreamcolor), but I still wonder if anyone knows if CS6 or other NLE can deliver a YUV to RGB accurate signal to an accurate monitor and you can trust it. Apparently CS6 is able to do that (Speedgrade?, AE?). Don't know about avid. It's not that I want to avoid buying an I/O card, but looking in the forthcoming 2-3 years that I won't upgrade my computer it'll be nice to know if we really need it for NLE/Color correction apps that the hardware don't support it.

Boo. Eyeball compared ipad mini and regular by pulling up the same color chart on both via google image search. Colors definitely dont look the same; mini's desaturated/washed out compared to full size. Maybe thathats only the new full size ones with the retina display that look that different; no way to tell (i did this at microcenter).

Researching online confirms, saying that mini displays 62% of full color gamut.

My understanding was that even if an HDMI monitor receives a YUV signal, most will still convert to RGB before displaying anyway.

Right! The CRT and LCD display technologies used in all televisions and computer monitors create color using red, green and blue additive primaries. If you look with a magnifying glass at any screen you will see these primary colors, either as dots or vertical stripes. The choice reflects human physiology: the eye has three types of cones that are sensitive to three different wavelengths of light--red, green and blue. Therefore, at some point, YUV must be converted to RGB in order for us to see the colors.

Accurate video monitoring involves calibration and correct viewing environment. Calibration is a result of software, firmware and hardware working together in a predictable way. A trivial mistake is when the difference between full and studio swing levels create what is sometimes called a gamma shift. PC systems can be configured in so many ways that it can be complicated to ensure an BT709 transfer function when displaying video. Macintosh computers are less configurable and therefore easier to calibrate. However, there is almost nothing to configure with a reference monitor directly fed by a YUV signal.

Correct viewing environment is as important as calibration. Computers monitors are generally viewed in well lit rooms while televisions are viewed in darker rooms. In addition, any windows open on your PC desktop, especially windows with black on white text, give off ambient light that affects the perceived brightness and color of the video window. This is why a dark color scheme defaults with most video editing software. Make the video window full screen and pay attention to room lighting.

In the end, your viewers will watch the video using a mixture of CRT televisions, gas plasma displays, various types of LCDs, computers and projectors. Many televisions will still have the vivid picture profile selected from the showroom at a discount store. For the same reason an audio engineer checks how popular music will sound in an automobile, most video should be checked on a variety of display devices. With experience, monitor calibration and attention to viewing environment one gets better at achieving reasonable image levels the first time around.

I'm thinking of buying an HP Dreamcolor (that can work in HD 709 color space) and connect it directly with a displayport cable. Like most people on this thread I've read the provideo coalition post, and if CS6 can really do an accurate YUV to RGB conversion, then I don't see the need for an I/O (for me). I just want an accurate broadcast preview. Does anyone know if this it's true?



You don't need an I/O card only for display. A 10-bit GPU is good enough, like Firepro and Quadro. Displayport is perfectly fine.

And by the way, the conversions from Y'CbCr to RGB and back are clearly defined mathematically (http://en.wikipedia.org/wiki/YUV#Converting_between_Y.27UV_and_RGB), with interpretations left to the user (you). If a manufacturer cannot do it correctly, they've made a simple math mistake. I have yet to see this in reality. If you're having color problems, or have heard of problems by other people, then something else was wrong.

Most consumer grade monitors can display 90% or more of Rec. 709 (sRGB), but a broadcast monitor should be capable of 100% - otherwise what's the point of spending extra?

I thought I read that the Dreamcolor needed a specific something something kind of input to activate its something something engine, otherwise you wouldn't be taking advantage of the color accuracy it's capable of.

What consumer monitors are you referring to? I would happily take 90% accuracy over 0%, which is what I have now.

I thought I read that the Dreamcolor needed a specific something something kind of input to activate its something something engine, otherwise you wouldn't be taking advantage of the color accuracy it's capable of.

What consumer monitors are you referring to? I would happily take 90% accuracy over 0%, which is what I have now.

Ha Ha! 0% as in switched off? You just need to switch it on.

If you're serious, the ASUS PB278Q and Ultrasharp series (the cheaper ones) easily reach 100% sRGB, which is about 72% NTSC. Most monitors above $1,000 should do 100% sRGB and Rec. 709 easily. This is more than enough for broadcast work HDTV mind you, not NTSC or PAL.

The monitors more than $1,000 aim to get 95% or more of Adobe RGB, which is wide gamut. If you're going cinema quality, you'll need a P3 based projector. The 'cheaper' cinema monitors are close, but not good enough. By cheap, I mean the most expensive ones by Eizo, NEC, Dolby, etc.

No, the Dreamcolor doesn't need anything extra. I confirmed that two years ago with HP and Nvidia. All it needs is a GPU that delivers 10-bit. Unless they lied...

Wish you and everyone else a happy new year!

My mistake on the dreamcolor, then. Maybe I'm thinking of what I read/heard about why it didn't make a good FIELD monitor.

I was being somewhat facetious about 0% accuracy, but I really am using an SD 13" CRT Sony production monitor hooked up via firewire and calibrated to color bars from my NLE to judge color at this time, which is wrong on many levels.

When you distinguish between NTSC and REC 709 (e.g. this monitor is good for REC 709, but not NTSC), are you talking about SD vs HD?

This would be for use on my personal projects (I don't offer myself professionally as a colorist) to myself piece of mind that on those rare occasions when I have a screening (festival, event etc.) or something, I'm sending out an image that looks how I think it does. So not broadcast, not cinema quality.

Everything I've read says that any computer monitor, no matter how nice, unless it's able to be calibrated to the 709 colorspace, is no good for that type of color correction. So that would be your Eizo, dreamcolor, etc. The rest are good for the RGB colorspace only (i.e. computer/web stuff). So now we're left with the $2500+ monitors like TV Logic, Flanders, etc., or some plasmas. And even those monitors are no good unless professionally calibrated (i.e. by someone who does it professionally for a living, not by you or me).

Did I get all that right?

When you distinguish between NTSC and REC 709 (e.g. this monitor is good for REC 709, but not NTSC), are you talking about SD vs HD?

Either way. You could assign an HD video to either PAL or NTSC if you wanted to. It sounds weird, but there are scenarios where it's useful. E.g., I am not very experienced with NTSC, but Rec. 709 is very close to PAL, and I remember moving between the two, temporarily, just to make sure my grade would work for both HD and the DVD version. You can't do that with NTSC, though.

NTSC is a better color space overall, compared to Rec. 709, sRGB or PAL. It covers roughly 54% of CIE XYZ, while PAL is about 37% and the other two are 35%. And cinema quality, either P3 or ACES, is about 70%. Don't ask me how I came up with these numbers. It's ugly.

Most monitors are either HDTV or sRGB, and the old broadcast field monitors are still around only because a lot of people are still shooting SD, instead of shooting in HD and down-converting it in post. When you're in such a production, having a monitor with the ability to do both is quite beneficial. But it's getting rarer by the second.


This would be for use on my personal projects (I don't offer myself professionally as a colorist) to myself piece of mind that on those rare occasions when I have a screening (festival, event etc.) or something, I'm sending out an image that looks how I think it does. So not broadcast, not cinema quality.

Everything I've read says that any computer monitor, no matter how nice, unless it's able to be calibrated to the 709 colorspace, is no good for that type of color correction. So that would be your Eizo, dreamcolor, etc. The rest are good for the RGB colorspace only (i.e. computer/web stuff). So now we're left with the $2500+ monitors like TV Logic, Flanders, etc., or some plasmas. And even those monitors are no good unless professionally calibrated (i.e. by someone who does it professionally for a living, not by you or me).

Did I get all that right?

A computer monitor capable of 100% sRGB is good enough for professional, broadcast quality HDTV grading and testing. I have written an article that might shed more light on this: What is Color Gamut and a Wide Gamut Monitor? (http://wolfcrow.com/blog/what-is-color-gamut-and-a-wide-gamut-monitor/)

But hold on, we are only comparing color space, and nothing else. A high-end Dell ultrasharp is the perfect color space, but it is a consumer device. An Eizo (the higher end ones, the cheaper ones are priced like the Ultrasharp) has internal calibration and are built for heavy, demanding work for years, just like the Dreamcolor. Unlike cheaper monitors or television panels, it will not lose quality. Not to mention they are 10-bit, while the consumer monitors are 8-bit.

For small production houses and content producers like you and me, we don't need anything better than a $2,500 Eizo, if it is paying for itself. If it isn't, we are perfectly okay with an Ultrasharp, or any monitor that does 100% sRGB at more than 350 nits or so.

Thanks. Now with these "good enough" monitors (a $2000 anything monitor, in my case is definitely not paying for itself), how does one calibrate them? A specialist? Spyder (or other similar device0? Via the matrox/aja/whatever i/o box one uses to send the video signal from one's NLE/CC software?

All you need is a good GPU and a calibration kit like the X-rite. Spyder's fine too.

Don't forget middle gray walls! You'll need a hood to cut off light hitting your monitor. Light reduces contrast and kills Dynamic range.

This throw any light on the Ultrasharps?


CS

PS: That any better?

I thought the idea was NOT to shed light on the Ultrasharp! GET IT? SHED LIGHT?????

But seriously. . .do you have that a little larger? It's very tiny on my screen.

Upload updated. Got it as a .gif too if that's more usefull.


CS

Thanks. Not sure what all that means, exactly. I'm guessing it's good.

Well, I'm sure Sareesh can tell us.

Don't pretend it means anything to me either, but they're beutifull screens.


CS

Sareesh melted my brain. I'm still reading that if you intend folks to see stuff in 709, that's the space you should grade in. Which brings us to the issue of LUTs for monitors in other color spaces. . .so if working with an sRGB monitor, how do you apply those? Is it through the CC software or an i/o box or something else?

Sareesh melted my brain...

Well, I'm sure Sareesh can tell us.

All right, you asked for it!

In traditional science, whenever you see the term 'Delta', it means 'difference' or 'deviation'. If I have seven cups to my left and three to my right, my cup-delta is 4.

In the image Chris has scared us with, Delta-E shows how each color deviates from the 'perfect color space' value. Every monitor, even the most expensive ones in the world cannot reproduce a space perfectly.

Color space is theoretical. The color space of the human eye is CIE XYZ, but even the eye cannot see all the colors in this space!

The Delta-E of the human eye is about 1. The best color grading monitors should have a Delta-E of less than 3. The Delta-E of the Ultrasharp is about 4, which they represent as <5 for both sRGB and Adobe RGB.

Delta-E has many variations, this one was ratified in 1994 by CIE so it's called DeltaE94.

So bottom line, the graph shows us that the Ultrasharp is better than consumer grade monitors, but not as good as the best monitors on the market - but we already know this, right?

The bottom-left graph should be perfectly horizontal, but it is not - I don't think there is a monitor capable of it, so I guess it's pretty good that it manages to be horizontal most of the way.

The bottom-right is just a gamma curve - at 2.2, which is standard. No vodoo here. It does tell you that the display max is at 250 nits (cd/m squared).

Bottom bottom line? Take all this with a grain of salt. Color graphs are to monitors as sensor megapixels are to camera manufacturers. Don't believe everything you see.


I'm still reading that if you intend folks to see stuff in 709, that's the space you should grade in. Which brings us to the issue of LUTs for monitors in other color spaces. . .so if working with an sRGB monitor, how do you apply those? Is it through the CC software or an i/o box or something else?

Yes, nobody claims any different. Walk into any broadcast studio and all you'll see are monitors in Rec. 709 (assuming they're shooting and broadcasting HD).

But, sRGB = Rec. 709, which makes any monitor that can show 100% sRGB good enough for Rec. 709.

LUTS can be applied two ways: software or hardware.

The software LUT tells the graphics driver/software to skew the data to make things look different. The monitor does not change.

The hardware LUT does the same via firmware living inside the monitor. The panel does not change.

In your case, Josh, LUTs are handled by software. That's good enough, and is the most versatile way to do things nowadays. The cool thing about LUTs is that they don't need a lot of computing power. It's like having a multiplication table in your hand so you don't need to calculate for yourself.

You can create your own LUTs, dump them in the software of your choice - whether it be color grading or editing or vfx - and it just skews the data to show you the space you are working in. A monitor is like a butler. It takes orders from its master (the GPU software) and tries its best. A butler can only be pushed so much.

Working with color profiles and LUTs are not as easy as I'm making it out to be. There's a lot of stuff happening under the hood that you should be aware of. I highly recommend you start working with free software, like Resolve, Red Cine-X, etc. and try to solve one problem at a time. Working in Rec. 709 is the simplest workflow in video - it has been designed specifically for ease of use.

Is anyone still reading?

Gothcha! Still reading.

Anyway, I hate to be "that guy" (confession: I'm ALWAYS "that guy"), but research indicates that while the primaries and white point are identical between sRGB and REC 709, the gamma is different, one is 2.2 while the other is 2.4 (guessing sRGB is 2.2).

So that means we need a way to tell the thing to display with a 2.4 gamma, and can basically leave everything else the same, do our Spyder calibration, and we're good to go, yes? So is altering the gamma a simple thing?


As for getting complicated, no worries; again, I'm not trying to be pro colorist. I just graded (it was a very gentle, conservative grade) my last short with the CC tools in FCP 6, and would do it again if I had to. I now have access to the tools in After Effects and Premiere (though someone pointed out AE doesn't have video scopes, at least not free ones. Boo.), and will start playing with the free version of Resolve soon.

sRGB and Rec 709 are equal color spaces. Gamma is a different thing altogether.

Gamma has nothing to do with color, it is applied on the luminance (Luma) only. When you change the 'color space' in software, the gamma is one of the things which are 'skewed'.

That's because Rec. 709 is a color space, but when used in broadcast it also includes many other things, like resolution, frame rate, gamma, etc. Just like NTSC, which is an umbrella term for color space as well as 59.94i, color, its gamma, etc. Professional software allows you to define gamma independently to color space. Consumer apps do not - they just provide 'presets'.

All this is nothing - the difference between 2.2 and 2.35 (the broadcast standard), or 2.4 (the rounded value) is not beyond the capability of any monitor, consumer or professional. A bigger jump is a gamma of 1.8, which Apple displays had pre-2009, and that caused a lot of problems.

Both sRGB and Rec. 709 are color spaces designed from the ground up to work equal to one another. They knew that one day HDTVs and computer monitors will probably use the same technology. Hasn't that turned out to be true today?

To answer your question: Altering gamma is a simple thing. Especially if you're working in professional software like Adobe's or Apple's or whatever. It can take care of everything under the hood so you can keep working, or it gives you the power to make your own variations, through color profiles or LUTs.

The GPU software just takes orders from your app and skews the data and passes it on to the display.

If your display is capable of 100% sRGB, and is calibrated for it - it is more than capable of working in the Rec. 709 color space for all practical purposes.

I just burned a 100 calories...