DreamColor disappointment: IT DITHERS!

Discussion in 'Displays' started by ToastyX, Nov 18, 2008.

  1. ToastyX

    ToastyX [H]ard|Gawd

    Messages:
    1,389
    Joined:
    Mar 12, 2005
    Dithering

    Not only is there dithering, the dithering is not that good. I can clearly see two different types of dithering being used: spatial dithering and temporal dithering using frame rate control. The dithering is most noticeable on dark colors, like on this page: http://www.lagom.nl/lcd-test/black.php

    The firmware can be updated by the user, but only in Windows. Out of the box with the original firmware, the spatial dithering looked like a fixed pattern of colored noise. After updating to the latest firmware, the spatial dithering looked like simple 2x2 ordered dithering. With both firmware revisions, the temporal dithering looked like faint noise scrolling up or down, and some shades had rolling diagonal lines.

    Here is an emulation of the spatial dithering before the firmware update:
    [​IMG]
    I kid you not. It really was that bad. Keep in mind that this is not an exact representation, but it's a good example of what I saw.

    Here is an emulation of the spatial dithering after the firmware update:
    [​IMG]
    If you don't see anything wrong, look closer. While the new dithering is much better, I could still see it. On top of that, there was still frame rate control, which added noise and rolling lines to some shades.

    I really hate to trash this monitor based on the dithering because other than that, it is a nice monitor, but a $3500 color-critical monitor should not have visible dithering, especially when the main selling point is 30-bit color with no dithering or frame rate control: "A full 30-bit pixel is sent from the DreamColor Engine to be displayed on the HP 30-bit LCD panel with no dithering or frame rate control." Source: http://h20202.www2.hp.com/Hpsub/downloads/DreamColor_and_LP2480zx_FAQ_June08a.pdf

    The dithering is definitely coming from the DreamColor Engine. 1080i over HDMI disables the DreamColor Engine, and there's no dithering there. The dithering is present in all other modes on both DVI and HDMI. I did not have a way to test DisplayPort, but that shouldn't matter. A 24-bit source on a 30-bit monitor should not have dithering either.

    It's possible that the NEC monitors (LCD2490WUXi and LCD2690WUXi) also have dithering, but if that's the case, I can't tell. On the HP, I can clearly see it.



    Panel

    According to the service menu and the TCO'99 compliance certificate, this monitor has an LG LM240WU5 panel.

    Service menu:
    [​IMG]

    TCO'99 compliance certificate:
    http://h20195.www2.hp.com/V2/GetDocument.aspx?docid=0900a5a582f9ad2d&cc=hr&lc=hr



    Uniformity

    I took white and black luminance measurements on nine parts of the screen:
    [​IMG]
    I did not see any major uniformity problems, but it's not perfect. This monitor does not have uniformity compensation like the NEC monitors do.

    I did not see any backlight bleeding at all:
    [​IMG]
    The white dot in the corner is the mouse cursor.



    Viewing Angles

    This monitor definitely has the A-TW polarizer, which prevents black from blooming at slight angles. It has the same faint red and green glow that the NEC monitors have.

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    The A-TW polarizer causes a red glow when looking from the right at extreme angles, which is why there's a slight red tint in that last picture. Based on my experience with the A-TW polarizer, some units will have the red tint from both sides, probably depending on the direction the polarizer was put on. There's no gamma shifting at any angle, as you can see.



    Color Spaces

    This monitor has the ability to emulate color spaces and provides seven color space options:

    FULL
    AdobeRGB
    Rec. 601
    sRGB
    Rec. 709
    DCI P3 Emulation
    User-7

    FULL has very intense reds. Think hot pink, but red instead.

    AdobeRGB looks very good to me. AdobeRGB has the same red and blue primaries as sRGB, so reds aren't overly intense. I've often recommended against wider gamut monitors for desktop use because then the colors are oversaturated, but that's mostly due to wider gamut monitors having overly intense reds.

    sRGB still doesn't look right to me. The reds are too pure and dark. I have never seen a perfect sRGB reference monitor, so it could be right for all I know, but sRGB is supposed to represent a typical monitor's gamut, yet the reds look nothing like any monitor I've seen. The NEC LCD2490WUXi better represents a typical monitor's gamut. Strangely enough, AdobeRGB has reds closer to what I expect even though sRGB is supposed to have the same red primary.

    Rec. 709 has the same problem with reds as sRGB, which isn't surprising since they both share the same gamut.

    I'm not familiar enough with the other color spaces to comment on them. They are used for video production.

    User-7 can only be set using the calibration kit, which I didn't have.

    The monitor only gives you control over luminance and color temperature. The gamma can only be set with the calibration kit.

    All the presets can be calibrated using the calibration kit:
    [​IMG]

    It looks like the monitor can be calibrated to any gamut, but I did not see options for esoteric gamma curves like sRGB.

    The software will only work with the colorimeter that comes with the calibration kit, so I couldn't test it.



    Brightness and Contrast

    FULL:
    White = 236.750, 236.668
    Black = 0.238, 0.237
    Contrast = 995:1, 999:1

    AdobeRGB:
    White = 162.831, 162.820
    Black = 0.173, 0.172
    Contrast = 941:1, 947:1

    AdobeRGB, lowest brightness:
    White = 44.978, 44.599
    Black = 0.048, 0.048
    Contrast = 937:1, 929:1

    sRGB:
    White = 81.908, 81.820
    Black = 0.101, 0.099
    Contrast = 811:1, 826:1

    This monitor can go down to a very low brightness without panel blocking. This is how all LCD monitors should be. The brightness ranges from around 45 cd/m² to around 235 cd/m².

    The white point is also set in the backlight itself without panel blocking. The RGB LEDs can be seen adjusting the white point when the brightness is changed. White looks very pure regardless of the brightness.

    The native contrast is also relatively high. This combined with the A-TW polarizer gives this monitor the best black I've ever seen in an IPS panel, but "CRT-class black" is a bit of an exaggeration. It's more like PVA-class black. This is the first time I've seen an IPS panel come anywhere close to 1000:1 contrast.



    Scaling and Video Modes

    Scaling options:

    Fill to Screen
    Fill to Aspect Ratio
    One-to-one
    Overscan

    The overscan option is useless. It cuts off five pixels from all sides, then scales to fill the screen. Why would anyone want to do that? There is no way to have overscan while preserving the aspect ratio, and five pixels is often not enough for television broadcasts.

    Action shot:
    [​IMG]

    1080p and 720p are scaled perfectly without overscan.
    1080i over DVI does some sort of bob deinterlacing, and it looks like it does something to the image before that, like maybe converting to 960x540 first.
    1080i over HDMI also looks that way but with better deinterlacing that doesn't bob.
    480p is treated as 3:2. Yet another monitor that can't scale 480p properly.
    480i doesn't work over DVI, and I don't remember testing it over HDMI.

    None of the modes are scaled properly over component. All of the modes have 2.5% overscan, and "Fill to Aspect Ratio" doesn't scale correctly, causing the image to be too wide.

    The DreamColor Engine is disabled for the analog inputs and interlacted modes over HDMI, which means you're stuck with the excessively wide gamut in those cases. There is a saturation control, but there's only so much you can do with that.

    DVI and HDMI are always full range, regardless of the color space.

    This monitor is a bust for anything other than 1080p and 720p full range RGB over DVI and HDMI. This monitor was obviously not designed for video.



    Lag

    I tested this monitor against the DoubleSight DS-263N. I could not get a stable image on the HP over DVI when connected to a DVI splitter, but the results were usable. HDMI provided a more stable image and produced the same results.

    DVI (left = HP, right = DoubleSight):
    [​IMG]

    HDMI (left = HP, right = DoubleSight):
    [​IMG]

    Every shot showed that the HP is exactly one frame behind the DoubleSight over both DVI and HDMI, which puts the lag at a little over one frame compared to a CRT (20-25 ms). Overdrive made no difference. This is consistent with what people have been getting on the HP LP2475w, so they both probably have the same amount of lag. It's not as bad as the NEC monitors, but I could still feel it. I doubt most people would notice it if there are no other lags present, but I wouldn't recommend this monitor for anything latency sensitive.



    Response Time

    The response reminds me of S-PVA panels. There is much less motion blur than I've seen on other IPS panels, but there's this effect, like I'm always seeing at least two frames on the screen at once. Overdrive didn't make a significant difference on that effect. I've only seen that in S-PVA panels until now.

    Overdrive can be turned on and off, but with it enabled, I kept encountering subtle artifacts in static images.

    For example, this image causes a faint flickering dot to appear in one of the four copies, depending on where it is on the screen:
    [​IMG]

    This is the first time I've encountered a monitor where overdrive produces artifacts on static images.



    Flicker

    This monitor has a very subtle flicker that I can see when I move my eyes, sort of like a DLP, except not as strong. I don't think most people would notice it, and I was not bothered by it. It's not like a CRT at 60 Hz or anything. When viewed with a camera, the viewfinder shows the monitor flickers from top to bottom. I wonder if they incorporated some sort of backlight scanning to improve the perceived response time.



    Coating

    The sparkly coating is back, sort of. It doesn't seem to be as bad as the older S-IPS panels, but it's not as smooth as the NEC monitors or the Planar and DoubleSight monitors.



    Noise

    I did not hear any buzzing when I put my ear up to the back unlike with most CCFL monitors, but I did hear a fan, which can be heard from a normal viewing position in a very quiet room.



    Unpacking and monitor pictures

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    There's not enough room to rotate the monitor without tilting it back first. Why was it designed like that?



    Defects

    I didn't see any dead or stuck pixels, but I did spot a mark embedded in the panel:
    [​IMG]



    Conclusion

    I returned it for the mark embedded in the panel, and I decided not to get a replacement due to the dithering. If it weren't for the dithering, I'd say it's one of the best monitors out there, but I don't know what to think of it now.

    I think most people would be better served by the NEC monitors. They cost half as much and perform nearly as well, without visible dithering. The LCD2490WUXi is good for sRGB and HDTV material, and the LCD2690WUXi is good for AdobeRGB material.

    The most unique feature of the HP is the ability to emulate color spaces, but not many people need that, and I don't know how accurate it is since sRGB doesn't look right to me.

    The HP can only be internally calibrated using the $350 calibration kit, bringing up the total cost.

    The HP has better contrast due to being able to lower the brightness and change the white point without panel blocking, but that only makes a minor difference. Most of the magic comes from the A-TW polarizer, which the NEC monitors also have.

    The HP has more inputs, but the analog inputs aren't even scaled properly, so what's the point? You can get one of the NEC monitors along with the Blackmagic Intensity Pro for component inputs and still have money left over. There are also plenty of component to VGA transcoders available. The DVI ports can be used as HDMI ports with a simple HDMI-DVI cable.

    The HP has slightly less lag than the NEC monitors, but not less enough to justify the price. It's still over a frame behind a CRT. If you care about lag, get the Planar PX2611W or the 23" Apple Cinema Display.
     
  2. DarkScythe

    DarkScythe Limp Gawd

    Messages:
    367
    Joined:
    Mar 7, 2008
    Thanks for doing a review of this thing, it was really damn difficult to find one that wasn't a 2-paragraph news brief.

    I don't really know how it works, but could the dithering have been a defect of your LCD, or is it something that would likely be a problem on all of them?

    I was really looking forward to this LCD as it seemed to be sRGB capable with a lot of inputs for my PS2 and such, but I guess the inputs don't work as well as the LP2475w?

    One thing that bugs me about the NEC2490 is that it seems to only have DVI and VGA inputs, and I'm not sure if it does PIP or input switching or all that other fancy stuff HP packed into the LP2475w. If only the latter was sRGB instead of wide gamut.

    I was going to ask if you could measure the color levels of its sRGB mode without doing any calibration, but I guess you can't since you've returned it. Nonetheless, thank you for those other measurements and tests.
     
  3. 10e

    10e 2[H]4U

    Messages:
    3,370
    Joined:
    Jul 20, 2006
    Wow, thanks for the info Toasty.

    Pretty disappointing for a $3k+ monitor especially in regards to the dithering. It looks like it first had temporal dithering and went to FRC type dithering after upgrading. Did both sets of patterns move? Looks like marketing over function on that one. Better to NOT advertise 1.04 gazillion colors and just not dither like that. I can imagine the havoc it will cause the heavy duty photogs out there.

    I have looked a lot at the NEC LCD26 and the Dell 2709W and neither has this type of dithering in close up photos or in real life (and the stupid Dell claims to have that 1.04/30-bit color). The top one looks like the BenQ FP241VW dithering and the bottom like any TN panel out there.

    Darkscythe, the NEC doesn't do PIP. If only the 2475W had a good sRGB mode. Most new wide gamut monitors have garbage sRGB modes.
     
  4. albovin

    albovin [H]ard|Gawd

    Messages:
    1,653
    Joined:
    Feb 16, 2008
    In one word: HP DreamColor is a big gimmick.
    Thank you ToastyX.
     
  5. richard13

    richard13 n00bie

    Messages:
    2
    Joined:
    Oct 21, 2008
    Albovin:

    "HP DreamColor is a big gimmick" is SIX words my friend... Six (6) words

    keep practicing.

    AND by the way, Albovin, I would like to thank you personally for helping me in another thread clear up several confusions of mine as well as for the excellent help in my monitor quest..
    BTW I had a great chat with William Hollingworth of NEC today , a very knowledgeable and excellent person, and will get a chance (finally) to see the 24 and 30 inch NEC monitors that is if all works well, at near by trade show.
    I will hopefully also get a chance to see the Eizo 301 by the end of the week...
    ToastyX?
    not a good review, but a GREAT review ... brief, to the point , even handed
    My profound compliments ..
     
  6. Yelnats

    Yelnats Limp Gawd

    Messages:
    228
    Joined:
    Jun 4, 2006
    Are all LED displays going to have some flicker(at a human perceptible rate)? How about OLED screens? I've only seen small OLED screens, and they all flicker.

    Right now, I'm just as annoyed by pixel-walk. My last three monitors have a significant amount, and my understanding of it is that a lot of the inversion problem could be easily fixed by the manufacturer.
    (I had to replace a screen on my Nintendo DS, and accidentally came across a screw that could be tightened/loosened to adjust the screen inversion/voltages).
    Did this display have any inversion problems?
     
  7. anonycorp

    anonycorp n00bie

    Messages:
    47
    Joined:
    Nov 18, 2008
    Hell of a review[​IMG][​IMG][​IMG] I guess it's time to look at different monitors.
     
  8. silent-circuit

    silent-circuit [H]ardForum Junkie

    Messages:
    16,229
    Joined:
    Sep 18, 2005
    Just out of curiosity, what's the image in your "action shot" from? Wipeout HD?
     
  9. ToastyX

    ToastyX [H]ard|Gawd

    Messages:
    1,389
    Joined:
    Mar 12, 2005
    No, the examples I posted were of the spatial dithering only. The patterns were static and did not move. On top of that, there was also temporal dithering using frame rate control, which added faint moving noise and rolling diagonal lines, kind of like what you'd see on a TN panel, except not as strong.



    I thought about that, but after seeing the firmware update change the dithering to 2x2 ordered dithering, it seemed very deliberate to me. I don't see how a defect can accidentally produce something that requires logic to implement. The fact that the dithering doesn't appear in cases where the DreamColor Engine is disabled also makes it seem intentional.

    I don't know how well the inputs work on the LP2475w, but I wouldn't be surprised if the component inputs have the same problem.

    The NEC monitors don't have PIP, but they do have input switching. That's another thing. Switching inputs on the HP requires pressing the input button twice, then selecting the input you want and pressing the select button. Switching from DVI-1 to HDMI requires five button presses involving three different buttons. The NEC monitors only require pressing a single button.

    I don't have a way to do that anyway, but I still have the raw data.

    The monitor seemed to be factory calibrated since the gamma curves required very little correction. AdobeRGB was very close to 2.2, and sRGB was definitely using the sRGB gamma curve.

    I'm not sure how accurate the white points or the primaries were since I don't have a colorimeter with filters specifically designed for this monitor, and I don't have a spectrophotometer either since those are expensive. The gamut changed with the color space, but the measured primaries didn't quite match the standards. I'm not sure how far off the measurements were. The calibration kit would have been helpful here.



    I didn't see any significant problems with pixel walk. The temporal dithering was stronger than any inversion present.



    Yes.
     
  10. InToGraphics

    InToGraphics [H]Lite

    Messages:
    126
    Joined:
    Jul 1, 2008
    What graphic card and cables were you using ?
     
  11. ToastyX

    ToastyX [H]ard|Gawd

    Messages:
    1,389
    Joined:
    Mar 12, 2005
    It's irrelevant. I saw the dithering on three different machines: a Mac Pro, an old PC when I updated the firmware, and a PlayStation 3 over both DVI and HDMI. The dithering is being done by the monitor.
     
  12. InToGraphics

    InToGraphics [H]Lite

    Messages:
    126
    Joined:
    Jul 1, 2008
    DVI is 8-bits per colour in and out.
    Regarding HDMI, you seem pretty sure that your source, and source material are HDMI 1.3 compliant. Your target definitely is.
    Of those 3 you mentioned only the PS3 has HDMI 1.3 capability (without Dolby TrueHD en DTS HD; but that doesn't affect the picture.)

    I don't know of any mainstream video card for instance which is HDMI 1.3 compliant.
    To my knowledge the cards which come with HDMI out, are HDMI 1.1 compliant.
    Maybe the FireGL V7700 card with DisplayPort out is maybe DisplayPort 1.1 compliant (don't know), and then it should be able to do the 10-bits per colour, if fully and properly implemented.

    Your cable might be HDMI compliant (supporting single-link bandwidth of 340Mhz/10.2Gbps) but if your HDMI source, source material, and target are not, then at least one end of the cable will dither, sometimes both sides of the cable will dither.

    If you want to properly feed the LP2480zx, i.e. make use of most (the first 3 ?) of the new features that HDMI 1.3 supports, those features you've decided upon must be active in the source and active in the receiving piece of equipment. Both ends must have HDMI 1.3 transmission circuitry (and all devices inbetween must also have this, such as switches, splitters, video processors, etc.).

    The features are at both ends of the cable, the bandwith for those features is provided by the cable and the transmission circuitry at both ends.

    And that is only the hardware.

    HDMI 1.3 adds support to the HDMI standard for :
     
  13. ToastyX

    ToastyX [H]ard|Gawd

    Messages:
    1,389
    Joined:
    Mar 12, 2005
    None of that matters. A 24-bit source on a 30-bit display should not have visible dithering.
     
  14. Eric Carroll

    Eric Carroll n00bie

    Messages:
    8
    Joined:
    Nov 27, 2008
    ToastyX,

    HP is apparently now aware of this thread. Dan Bennet, from HP, posted in the CreativeCow forum about these issues and others. Check out the thread: http://forums.creativecow.net/thread/223/7928#7953

    There is also a lot of discussion going on about the Dreamcolor display on Reduser.net Off Topic forum.
     
  15. Snowdog

    Snowdog Pasty Nerd with Poor Cardio

    Messages:
    7,769
    Joined:
    Apr 22, 2006
    Interesting. This is supposed to have a real 10bit panel. It should be fantastic with no dithering. The NEC 90uxi series are really only 8 bit panels, they actually are known to do dithering to provide a 12bit range. But I guess they do it pretty well. I haven't noticed it.
     
  16. albovin

    albovin [H]ard|Gawd

    Messages:
    1,653
    Joined:
    Feb 16, 2008
    IMO it's better to stay far from products that require new firmware next day after their firework premiere.
     
  17. albovin

    albovin [H]ard|Gawd

    Messages:
    1,653
    Joined:
    Feb 16, 2008
    "I haven't noticed it" - this is the key phrase. It will last longer than HP with all possible firmware.
     
  18. albovin

    albovin [H]ard|Gawd

    Messages:
    1,653
    Joined:
    Feb 16, 2008
    I started reading:

    Highlighted in green - marketing from the fist line.
    Enough reading.
     
  19. Eric Carroll

    Eric Carroll n00bie

    Messages:
    8
    Joined:
    Nov 27, 2008
    You need to read a bit more.
    I have no stake in this other than trying to get the facts for my own buying decision. I neither endorse nor reject HP at this point.
     
  20. ToastyX

    ToastyX [H]ard|Gawd

    Messages:
    1,389
    Joined:
    Mar 12, 2005
    The new firmware he mentions is the version I tested when I updated the firmware, which was released in October. There isn't anything newer as of this post.

    Also, contrary to the manual and what he posted, the DreamColor Engine is enabled for 1080i over DVI but with a simpler deinterlacing method that bobs. Only 1080i (and probably 480i) over HDMI has it disabled.
     
  21. Eric Carroll

    Eric Carroll n00bie

    Messages:
    8
    Joined:
    Nov 27, 2008
    Ah, he said "The latest (as of today, 12/02/08) ", which is not the same as new firmware available today. Sorry, my mistake I will edit my post.
     
  22. InToGraphics

    InToGraphics [H]Lite

    Messages:
    126
    Joined:
    Jul 1, 2008
    Yes it does matter. What I'm saying is that a true and full 30-bit pipeline, eliminates any dithering. You do not have to wait for the next firmware release with upgraded dithering, because the firmware doesn't have to do any dithering with a full 30-bit pipeline.
    I have seen the difference with my own eyes. With and without I/O Box (which "dithers/scales" in hardware because the graphics card will be 8-bit),
    Stick to your 8-bits pipeline and in fact you're asking the firmware to replace one of these : http://www.matrox.com/video/en/solutions/devices/
    Or get yourself a nice AJA Xena or Cona card which comes with the proper plugins for your software, Premier Pro, After Effects, Photoshop, Combustion, Digital Fusion, etc.
    But that limits you to animation, effects etc.
    There is not much for still graphics editing in full 30-bit around, other than I/O Boxes and not even a handfull of graphics cards.

    When you buy a monitor like the DreamColor, be prepared to pay at least the same amount for additional hardware/software to get a full 30-bit pipeline and benefits.
    (Only if you didn't already have a true 30-bit pipeline setup.)
    HP should make this clear, when they say "the first affordable".

    The monitor is there, the hardware and software to drive it exists mainly in film and animation.
    This was predictable because you only have to look at who HP co-developed the monitor with.
    If you're not into film/effects/animation editing, or (pre-)press, look elsewhere or pay up.
    When you are in the press business, ask HP what else you will need to reap the full benefits of the DreamColor. They will tell you that you have to buy their software, printers, supplies, etc.

    Why take the easy route when you can take the hard route ? (this is not a mistake.)
    Why go full 30-bit when you can mismatch with "less than" hardware ?
    So become full 30-bit, or wait forever for the next "perfect" HP firmware release.
     
  23. philjohn

    philjohn Limp Gawd

    Messages:
    261
    Joined:
    Apr 1, 2008
    Interesting that he said when you give the monitor a 10 bit signal it will accept it, does it have a DisplayPort input in that case?
     
  24. Eric Carroll

    Eric Carroll n00bie

    Messages:
    8
    Joined:
    Nov 27, 2008
    The LP2480zx does support DisplayPort, HDMI 1.3 and dual link DVI interfaces all of which handle 10bit.
     
  25. Snowdog

    Snowdog Pasty Nerd with Poor Cardio

    Messages:
    7,769
    Joined:
    Apr 22, 2006
    There is more to it than a connector. You need a card that will work in 30 bit mode, you need software that will work in 30 bit mode...
     
  26. philjohn

    philjohn Limp Gawd

    Messages:
    261
    Joined:
    Apr 1, 2008
    True, but it's nice to know that when those are all in place it'll work. Are any cards 10bit capable at the moment (Quado FX for instance).
     
  27. Cyberbeing

    Cyberbeing Gawd

    Messages:
    557
    Joined:
    Nov 22, 2007
    Well considering the driver release notes, as of October this year it sounds like nVidia's DisplayPort cards (Quadro CX, Quadro FX 4800, and Quadro FX 5800) support 10bpc over DisplayPort in applications that support it.

    nVidia's Quadro SDI cards have supported 8bit, 10bit, and 12bit over SDI for ages.