ProVideo Coalition.com: Themed Channel | Editing & Post by Patrick Inhofer: HP’s DreamColor: A PVM CRT Replacement?

A last-minute job this past weekend seemed perfect for my testing. It was a 5-camera 1-hour comedy show, shot in 1080i… Perfect! The source material had lots of dark audience shots that needed brightening, peak whites that needed managing, the codec was DVCProHD (which can often have lots of compression artifacts in the shadows that I need to see and suppress), and I can test the various converter boxes to see how they handle feeding an HD-SDI interlaced signal into the RGB-progressive-inputs-only DreamColor.

The DreamColor I used was lent to me by David Bell and Lisa Yimm of HDR-VFX.com. The panel has 6,000 hours of usage, which gives me a good example of how the DreamColor will respond after almost two years of regular usage.

The HP DreamColor Colorimeter creates a profile in less than 5 minutes.

Setting Up the DreamColor

The first step is properly calibrating the DreamColor. This panel was built for RGB film workflows and has a color gamut exceeding both the the Rec 709 (HD) / Rec 601 (SD) broadcast specifications (it just misses the full Digital Cinema DCI-P3 spec, which this LCD will display in an ‘emulated’ mode pdf). If I ran the DreamColor at it’s full gamut, my colors would be super saturated - forcing me to heavily de-saturate the images. In HD, in SD and on the internet those resulting images would look very de-saturated and no one would ever hire me again!

So, I went out and bought HP’s Colorimeter (background info on this probe are in Allan Tépper’s piece). My first reaction was… horrific. The calibration had gone horribly wrong. After many hours of recalibrating on both the Mac and PC and then enlisting the assistance of HP’s very helpful DreamColor product team, it became clear that I had a defective colorimeter.

Running out of time and with no opportunity to exchange the defective unit, I managed to dial-in the DreamColor panel by using the 1D Lut builder inside the Decklink HDLinkPro DisplayPort. I got a near-perfect match on colors between the DreamColor and the Sony PVM monitor.

Grading with the DreamColor

In the first hour of my two day grading job I ran the PVM CRT and the DreamColor LCD side-by-side. I spent quite a bit of time working shadow detail, ensuring anything I saw on the CRT was also visible on the LCD. I was satisfied that the DreamColor was showing me everything I saw in the PVM. I turned off the CRT and spent the next 10 hours grading only with the DreamColor.

Monday morning, after having rendering out the concert overnight, I powered up the CRT and watched the show… without the DreamColor.

That moment, in my mind, is when the DreamColor proved itself every bit as capable as the Sony PVM 20L5.

There were only two things left to do: Get a properly functioning probe to see if I could get an accurate calibration. And test out NTSC footage.

Running the calibration software with the DreamColor attached to a Mac

Calibrating the DreamColor

I’m happy to report, a properly functioning DreamColor calibration probe works wonderfully. The software is easy to use. It runs on both a Mac and PC (I tested on both platforms). It also allows for calibrating to custom white points or color temperatures. Calibration does require a USB cable also hooked up between the host computer and the DreamColor (this is how the final profiles are uploaded onto the LCD itself).

There is one thing I need to point out that nags at the back of my mind… the calibration only works when hooking up the DreamColor directly to the host computer off the graphics card. I’d much prefer the DreamColor would be calibrated with the test images following the same signal path of the images feeding it while in production. In my case, I’d prefer the test images being fed out the Kona card into the Decklink or AJA converter boxes and then into the DreamColor.

NTSC On An LCD?

Top Image : NTSC image at 1:1
Bottom Image: NTSC image scaled
to fill the HP DreamColor

Understand this: In HDMI, DVI-D, or DisplayPort (the 10bit RGB Progressive inputs which engage the DreamColor Engine for gamut mapping) the panel is in full 1:1 mode. No scaling at all, thank you very much. This panel will show you what’s feeding it.

The Decklink and AJA converter boxes have options to automatically scale NTSC / PAL sources to fill the panel. These images are very clean considering the amount of scaling going on (the picture on the left demonstrates how small an NTSC frame is inside any HD monitor. In my book, that’s a big honking blow up.)

In Apple’s Color, things can get tricky. It’s entirely possible to set Color to handle the viewing upscale to 1080p. In this mode, viewing anamorphic DV can look pretty ugly. I found if I kept the entire pipeline outputting NTSC out of the Kona card and let the HDLinkPro / AJA boxes to handle the upscaling, the images were acceptable for evaluation.

My take on this issue: If you do lots of NTSC work, you need to evaluate this particular workflow to determine if upscaling is a problem for you. For me, it’s not an issue as long as I allow the HDLinkPro or AJA HDP-2 to handle the job.

Conclusions

How Bright, This DreamColor?

I want to dedicate some digital bytes on one issue I kept reading about on the various forums: Brightness.

Owners of the DreamColor seem to fall into two camps - those who think the brightness levels are just right and those who think it’s too dim. From what I can tell, those in the latter camp (too dim) tend to be PVM owners while those in the former camp (just right) tend to be BVM owners.

Remember my thoughts on BVM vs PVM monitors? And how my reactions run when moving between the classes of monitors? I think that’s what we’re seeing in the ‘brightness’ discussion. In the DreamColor calibration software, the end-user has to set the target luminance level for the panel. Industry accepted levels fall between 80 cd/m2 & 100 cd/m2, the spec calls for 120 cd/m2. The DreamColor is capable of up to 250 cd/m2.

In my testing, to match the brightness of my PVM - I had to push the DreamColor to around 165 cd/m2. Doing so, in my opinion, is a huge mistake.

The point of calibrating the DreamColor isn’t to get this LCD to match our aging PVMs, but to have a display that’s 10bit, has the full color gamut of Rec 709, and can be externally calibrated to match the DreamColor’s profile to that of the Rec. 709 specification.

Pushing the DreamColor to 175 cd/m2 is similar to buying a plasma for the edit room and setting it to the “Vivid” mode and color correcting to that. I’m sure the image will look great in that edit room - but it has nothing to do with giving your clients a properly calibrated panel to make informed accurate decisions about the image… decisions that will match thousands of other industry professionals doing the same thing on thousands of hours of programming… all (hopefully) using monitors also matched to the same specification.

What specific luminance level should we target? I consulted several people on this question including Greg Staten, former Product Design for the Avid Symphony, Author and HP’s DreamColor Solutions Architect, and Alexis Van Hurkman, a colorist and author who’s a colleague and has a chapter in his upcoming book on this issue. Their answers both jibe with my general understanding:

The specification calls for the room’s ambient lighting to be 10% of the peak luminance value (peak luma) of the monitor, with peak luma (per the Rec 709 spec) being set at 120 cd/m2. Most edit rooms have ambient lighting darker than what the spec calls for, so most of us lower the peak luminance of our monitors to somewhere between 100 cd/m2 and 80 cd/m2. Greg from HP suggests not running above 120 cd/m2 for HD and SD work. I finally settled on 100 cd/m2 as being appropriate for my room. If you keep your DreamColor running between 80 cd/m2 - 120 cd/m2, I think you’re safely within the Rec. 709 specs.

Additional Thoughts

I really like the HP LP-2480zx DreamColor. Another colorist friend of mine dropped by after reading my Twitter and Facebook updates on my escapades torturing this LCD. His reaction?

The viewing angle is sufficiently wide for a typical edit room. In a large room, he thinks he’d place a second client-dedicated panel. He grades in a theater environment with a projector and said this panel feels very similar to his room. Like me, he felt shadow detail was every bit as good as the PVM CRT.

At one point we put the PVM and the DreamColor side-by-side. The colors tracked extremely well, though clearly the PVM was pumping a much brighter image. But we both agreed that we’d stick with the Rec 709 spec at 100 cd/m2. Once we turned off the CRT there was nothing about the LCD we didn’t like.

We did comment on how bright the panel is when displaying pure black, suggesting elevated black levels. But feed it an image and ‘apparent contrast’ kicks in with a satisfying image.

Final Conclusion

I think HP’s LP-2480zx DreamColor is a terrific replacement to the industry standard PVM- series of Sony CRTs. To match the performance of those PVMs the DreamColor needs to be supplemented with an external probe. AND… the LP-2480zx has some very specific input requirements if you want the full benefit of the external probe. So you’ll need to follow Allan Tépper’s charts (here and here) on suitable devices to feed this monitor.

And please, please, please… don’t pump up the luminance levels to match that trusty PVM CRT. Retire the CRT to a less demanding offline suite. It’s not fair to the PVM, the DreamColor, or your clients to run both monitors in the same room.

I did find two shortcomings on this panel: The panel’s firmware can only be updated via a Windows machine and no Closed Caption decoding. I do quite a bit of Closed Caption encoding in FCP 7. The Sony PVM CRT has decoding built-in, allowing me to see my captions as I’m recording to tape. No such luck with the LP-2480zx. Bummer. Maybe in the next version? A finisher can dream, can’t he?

About The Author

Patrick Inhofer is the Owner/Operator of the boutique finishing house, Fini.tv and the training site TaoOfColor.com. He is a colorist, finisher and editor with over 21 years experience in broadcast, cable, corporate, and indie post-production. His passion for the industry was renewed with the 2006 release of Final Touch (now Apple’s Color). His color grading passion combined with his extensive background as an Online/Finishing Editor creates great left-brain/right-brain interaction that he believes is well suited to his temperament and the craft.

Patrick is also in the throes of launching a new color grading training website, TaoOfColor.com, featuring video training, forums, feedback, and mentoring.

Blackmagic: We’re ready to remove the Band-Aid!

Allan Tépper | 05/11

If you agree, please sign the online petition requesting the required updates.

Despite years of diplomatic prodding on my part, both via articles in ProVideo Coalition magazine and private emails, Blackmagic has still avoided and postponed offering RGB on its HDMI…

How the Blackmagic Cinema Camera will indirectly take sales from AJA, Matrox, and MOTU

Allan Tépper | 05/10

What are you talking about Allan? AJA, Matrox, and MOTU don’t manufacturer or sell cameras! How will the Blackmagic Cinema Camera take sales from AJA, Matrox, and MOTU? The reason is related…

Sony quietly announces the NX30 camcorder, a little sister to the NX70

Allan Tépper | 05/08

With an 1/2.88" sensor and 26mm wide angle (35mm eqv), the NX30 should ship in June for well under US$2500.

Although during the past year I’ve written quite a bit about the Sony NX70 (officially, the HXR-NX70) here in ProVideo Coalition magazine, I haven’t…

You must be registered to comment. This is an effort to reduce spam. Please REGISTER HERE.

Zak,

Thanks for your question.

I tend to stick with Decklink or Kona SDI cards, so I’m not very familiar with the MXO product line. However, it seems Allan Tépper already covered this base in an article on the MXO and DreamColor earlier this year.

No one can say the PVC doesn’t have the DreamColor thoroughly covered!

- patrick

Posted by Patrick Inhofer on 06/28 at 07:31 PM

Zak,

If you have read my four main DreamColor articles (of which Patrick has linked three in his article and the fourth in his first response to your question), then you already know that it is absolutely necessary to have some professional interface between your Mac and your DreamColor if you are using Apple software, like FCP or Color. The original MXO is really the only possible pro interface to use for that purpose if you are using a Mac that has no PCIe or ExpressCard34 slot. However, if you are using a Mac that does have either of those slots, then you are much better served with an MXO2 series interface box (as indicated in one of my charts), especially one with the MAX option factory installed inside. You will spend less and get much more.

Allan Tépper

Posted by Allan Tépper on 06/29 at 07:35 AM

Thanks for your pro insights in this review!

Interesting re the limitation of calibrating only when it’s connected to the graphics card.

I’ve been considering adopting this monitor but specifically to connect to an MXO2 device. My reasons are numerous and will be obvious to some and are generally outlined in my comments to Allan’s reviews on the monitor.

Along with the various selling points of the MXO2 solution, Matrox, as of NAB, has begun positioning it as an alternative to Nvidia’s Cuda GPU acceleration of CS5’s Mercury Engine.

Given that possibility, I was hopeful that the expense of a fancy graphic card to power Mercury was no longer required. If one can only calibrate Dreamcolor if connected to graphics card, it begs a couple of questions:

What card did you use?

I assume the fancier the better to properly calibrate. And yet, I’m puzzled about the fact that the Dreamcolor engine requires an RGB signal to operate properly and yet can be calibrated via a connection to a graphics card that doesn’t supply an RGB signal (none do as far as I know)

Thanks.

wsmith

Posted by wsmith on 06/29 at 08:59 AM

WSMith,

Graphic cards (whether they have VGA or DVI outputs) are by definition RGB. DisplayPort and miniDisplayPort could be either RGB or YUV, but so far have been RGB.

Allan Tépper

Posted by Allan Tépper on 06/29 at 09:07 AM

WSmith,

What Allan said!

It did occur to me this morning (while showering) that I could have tried running the LCD as an extended desktop via the Kona card. That might have worked and solved that one issue!

I don’t have the DreamColor in-house anymore. But when I do get one again, it’s the first thing I’m going to test.

My graphics card: ATI Radeon HD 3870

- patrick

Posted by Patrick Inhofer on 06/29 at 09:14 AM

Thanks Allan,

Yes of course, I somehow forgot that fact. (And they all output a progressive signal too).

This makes me wonder why an Nvidia Cuda card (required for accelerating MXO2) cannot be used to drive the Dreamcolor engine.

Of course it would need to have two output connections: one for the computer monitor and one for the external monitor. The Nvidia Quadro FX (also accelerates Mercury) has 2 outputs.

I guess I need to go back to your articles to refresh my memory but am I missing something on that?: Can we not use a Nvidia Quadro to power the Dreamcolor engine?

Thanks,

wsmith

Posted by wsmith on 06/29 at 09:35 AM

WSmith,

You can’t because (as I have covered in many articles) Apple’s realtime software conversion from YUV to RGB is not to be trusted for critical evaluation, as Apple has stated in Apple’s own bulletin.

Allan Tépper

Posted by Allan Tépper on 06/29 at 09:40 AM

WSMith,

Regarding calibration, the LP2480zx DreamColor display has a built-in 10-bit pattern generator and that generator is used to create the color patches used during the calibration process.

Only after calibration do we switch to the graphics card output to perform validation using patches generated by the graphics card.

Also, as Allan mentioned, every computer graphics card outputs RGB signals. Some graphics cards can also output YCbCr signals - typically via an HDMI connection - but they can always output RGB. Indeed, the vast majority of computer displays sold today can only accept an RGB signal.

Best,
-greg

Posted by .(JavaScript must be enabled to view this email address) on 06/29 at 09:48 AM

Hi Allan,

I’m on Windows, not Mac.

(All the better with respect to updating the monitor’s firmware too, apparently…)

So for Windows users, what about using an Nvidia Cuda card to power the Dreacolor engine?

Thanks,

wsmith

Posted by wsmith on 06/29 at 10:09 AM

wsmith:

I’m assuming you mean an NVIDIA workstation graphics card (the Quadro line)? They work perfectly and I’d bet that at least 80% of all LP2480zx DreamColor displays in use are connected to these types of cards.

Regarding using Cuda to power the DreamColor engine, the engine is internal to the display so it would not be possible to drive it externally.

Best,
-greg

Posted by .(JavaScript must be enabled to view this email address) on 06/29 at 10:15 AM

FYI: On Nvidia’s site I see the PDF below which in the 2nd sentence of 2nd paragraph of the intro, it specifically refers to the Dreamcolor.

http://www.nvidia.com/docs/IO/40049/TB-04701-001_v02_new.pdf

Posted by wsmith on 06/29 at 10:23 AM

WSmith,

I might have not been clear - once the calibration is done, the DreamColor doesn’t need the graphics card on the host computer. CUDA doesn’t come into play at all.

The DreamColor Engine is an on-board processor and the display handles it independently. A good thing, since it doesn’t eat up other processing cycles.

If you take a look at the links about the various ways to properly feed the DreamColor - Allan identifies which gear also work on in a PC pipeline.

Posted by Patrick Inhofer on 06/29 at 10:24 AM

Thanks Greg,

yes, understood re the Dreamcolor engine being internal to the monitor. And that it just needs a 10 bit, true progressive, RGB signal to work.

The Nvidia PDF apparently states (haven’t read all the details yet) that it outputs sRGB and Adobe RGB.

What can you say about those in terms of closeness to Rec 7009? I think I heard somewhere that one of those actually slightly exceeds Rec 709.

I’m a bit unclear on my RGBs (my external Sony LMD 2030W supports Rec 709 so I’ve just relied on that and I haven’t yet dabbled in serious color correction)

So how does sRGB or Adobe RGB fit into a Rec 709 monitor?

Thanks

wsmith

Posted by wsmith on 06/29 at 10:31 AM

Hi Patrick,

I pored over all of Allan’s prior articles with great interest and closeness when published and
I think I understand all his points.

I understand your point re calibration and via a graphics card but all of this still leads me to wonder why a Quadro cannot be used to drive to Dreamcolor operationally, not just for calibration?

Aside from the sRGB and the Adobe RGB output to the monitor.

Thanks,

wsmith

Posted by wsmith on 06/29 at 10:36 AM

Hi wsmith,

An NVIDIA graphics card can output virtually any color space required. Indeed, since we aren’t applying a LUT (lookup table) in the graphics card, it really doesn’t even care what color space is used in most cases.

Regarding the Rec. 709 color space, the primaries and white point are identical to the sRGB color space. AdobeRGB has a wider green primary and isn’t really relevant for television/film production work.

Best,
-greg

Posted by .(JavaScript must be enabled to view this email address) on 06/29 at 10:39 AM

WSmith,

Ah, I understand now:
“why a Quadro cannot be used to drive to Dreamcolor operationally, not just for calibration?”

I suppose you could. I know on a Mac, when driving a monitor from the graphics card the operating system’s color management system can’t apply two different profiles to two outputs on the same card. So I couldn’t apply one setting to the DreamColor and another setting to the second output which is calibrated to look like the DreamColor.

I don’t know if this is a limitation on the PC side. But even if it did, just so long you use the DreamColor as your reference monitor then you’re fine.

And the Quadro is passing through a 30bit signal, so that aspect of it works.

Greg or Allan, any reason why you couldn’t use the on-board graphics card to drive the monitor as wsmith is wondering?

- patrick

Posted by Patrick Inhofer on 06/29 at 10:54 AM

Patrick and WSmith,

The reason is for what I just wrote above. It depends upon the software, and is the topic of an article I am writing now. Stay tuned!

Allan Tépper

Posted by Allan Tépper on 06/29 at 11:10 AM

Great! Looking forward to it!

Thanks,

wsmith

Posted by wsmith on 06/29 at 11:18 AM

“I know on a Mac, when driving a monitor from the graphics card the operating system’s color management system can’t apply two different profiles to two outputs on the same card.”

That doesn’t apply for all Macs. I drive an HP L2335 as a 2nd monitor from both my 15” MBP (ATI Radeon x1600, OS X 10.5.8) and from my 2009 mini (nVidia 9400, OS X 10.6.3; primary display is a Sony S204) and I have no problem bringing up different profiles on both displays simultaneously and matching them as closely as their panels and backlights allow. The same was true when, instead of the mini, I had a PowerMac G5 with a Radeon card, and when I had the Dreamcolor at home, I was able to obtain a fairly good match with the other displays (though the Dreamcolor was and is clearly superior to the others), and this required radically different color profiles to be assigned to the two different output ports on the same GPU.

At work we have MacPros with ATI 4870 GPUs and OS X 10.5.8. On those machines, too, I’ve had no problem running different calibrations on two different displays.

Posted by Adam Wilt on 06/30 at 06:06 PM

Adam

I’m always happy to be corrected by you.

Thanks for the clarification!

- patrick

Posted by Patrick Inhofer on 06/30 at 06:49 PM

Hi Adam.

You’re absolutely correct. If you have an ATI graphics card you can apply separate output LUTs to each connected display. But, if you are using a current (or previous) generation NVIDIA graphics card you cannot. This is a limitation in the current NVIDIA architecture.

Note that if you are using an NVIDIA graphics card, the output LUT specified for the PRIMARY display will be applied to both displays connected to that card. The only way around this limitation is to use more than one NVIDIA graphics card (or to use an ATI graphics card, but that is not always an option).

Best,
-greg

Posted by .(JavaScript must be enabled to view this email address) on 07/01 at 07:51 AM

Hi Greg,

Surely that’s not a blanket limitation on all NVIDIA GPUs. My mini uses the NVIDIA GeForce 9400, and I have no trouble running different color profiles on my two different monitors.

Posted by Adam Wilt on 07/01 at 10:10 AM

Hi Adam.

I’ve primarily tested the Quadro FX line of graphics cards and it is a limitation on those. I’ve not tested the 9400 on the mini. Good to hear that it works on that one.

-greg

Posted by .(JavaScript must be enabled to view this email address) on 07/01 at 10:19 AM

Good to know about that limitation on the Quadros; we may be upgrading one or two workstations with them… fortunately we have dual Dreamcolors on them so the question of matching dissimilar monitors won’t come up (grin).

Posted by Adam Wilt on 07/01 at 10:39 AM

As promised, here is the article called: Does Premiere CS5 achieve the “impossible dream” for critical evaluation monitoring?

Allan Tépper

Posted by Allan Tépper on 07/02 at 07:15 AM

Its really lovely and i just would like to thanks to all Sony team members who really does creative works

Posted by MoviesBlaster on 07/10 at 03:55 PM

We got a Dreamcolor and the Advanced Profiling Solution. Unfortunately the X-rite is not seen by the software (but it is shown in the system profiler), so I can’t activate the program.
The HP hotline here is a bad joke.
Any idea what we can do?

TIA,

Uli

Posted by .(JavaScript must be enabled to view this email address) on 07/14 at 02:30 PM

You don’t say what platform you’re on, or which version of APS you’re running, so it’s hard to know what to say.

I’ve run APS 1.0.2 and 1.1.0 for Mac on various OS X versions across four machines in the past year, without problems. I’ve had the X-rite probe plugged into the Mac itself, or into the hub on the display; both have worked. (But if you’re not able to see the probe with the software, try a different USB port.)

When we’ve had problems (we have 1st-gen Dreamcolors, and we’re currently working through a variety of issues with them), I’ve had to spend an hour on the phone with first-level support just to get enough of a trouble ticket into the system to get a callback from incredibly helpful tech support folks in Ft. Collins. Once you get the real support folks online, they’re very, very good about solving (or at least attempting to solve) your problem.

This matches my experience with an older HP LP2335 display; initial problems were met with complete and total support from HP, once you can get past the first-line telephone-answering folks.

Posted by Adam Wilt on 07/14 at 04:55 PM

Hi Adam,

thanks for the info and sorry for not giving enough detail. Yes, it was under OS X 10.6.4 and APS was version 1.1 , tried two machines and different ports.

I’ll try to get past the first line of defense…

Best regards,

Uli

Posted by .(JavaScript must be enabled to view this email address) on 07/15 at 02:47 AM

It could also be a defective probe.

See if you can exchange it. MIght be quicker than fighting through their defenses!

And rather than fight through HP’s defenses, scroll up these comments to Greg Stanton and click on his name to send him an email… he IS the real HP DreamColor support guy that telephone support would be putting you in touch with.

- pi

Posted by Patrick Inhofer on 07/15 at 12:22 PM

Name:

Email:

Location:

URL:

Smileys

Remember my personal information

Notify me of follow-up comments?

Submit the word you see below: