Forums >
Photography Talk >
iPhone v4 color profiling
See here for a better (final) profile and writeup. ---------------------------------------------------- Yes, I got one early. Yes, it's awesome. OTOH, the color and contrast ratio are markedly different than previous generations. And as I've spent way too much time working on getting around this, I thought I would share my best existing solution. Alibis: - I tried to calibrate automatically using my Spyder3 and a VNC connection, but the lag time was too great. I've contacted Datacolor to see if they're planning a dedicated solution (doubtable) or can at least allow a timing adjustment in the software (fingers crossed). - As the above didn't work, this was done entirely by eyeballing it in a variety of lighting conditions. The only constant to that was that I attempted to match perceptual brightness between the iPhone and my calibrated reference screen, with the iPhone brightness set just above its midpoint. - My eyesight isn't the best, so corrective lenses may be leaving my perception with some bias. 'Solution': Here is a custom color profile which I created for the iPhone4 which attempts to remedy some of the color casts on my iPhone. As they're all supposed to be factory calibrated before leaving, this should at least be close to others, but I make no guarantee of it, nor that my idea of 'right' is the same as yours. To use it, simply install the profile on your system and in PS choose Edit->Convert to Profile, selecting "iPhone4v1" as the destination profile. There may be some shifts in the color on your screen, and that's normal. Save the file to wherever you keep your iPhone photo library, and sync up. (Hopefully) you'll notice a marked improvement in the color and contrast when viewed on the iPhone. Closing: The iPhone is not color managed, so understand that you're applying a sort of 'reverse' fix to the problem. As well, because you're converting to a new color profile and a new gamma, you'll get the best results making this change on a 16bit version of the image - whether you'll see that degree of difference in the result on the phone is another question. Again, I don't represent this as being anything more than what appears to be an improvement to me, on my iPhone. Your results may vary and you may think it looks like crap - that's fine. But no one else (which Google can find) is talking about this, it's color and portfolio representation relevant, and so I'm sharing it with you here . And now I'm going to go drink some coffee, because I'm exhausted. Edit: Here is a newer version which is a bit more accurate. Jun 24 10 05:12 am Link SRB Photo wrote: this would never work since the way the spyder works is to create a custom monitor profile that your Mac uses to adjust for gamma and white point. this custom monitor profile is NOT copied to your iPhone, so it is useless. Jun 24 10 06:13 am Link silverystars wrote: I'm well aware that the iPhone isn't color managed (sad, true). However, generating a profile of the display and then converting files for it to said profile will allow them to display correctly on it. Not anywhere else (well, if the profile is embedded and the other viewer is managed we're fine), but we're getting good color on the iPhone. And it's not much good as an in-a-pinch portfolio tool if the color sucks. silverystars wrote: And this is why I tested it in multiple light conditions and stated my brightness setting. It is a moving target, and the new screen isn't much better about color shifting with brightness changes, but it is an improvement. It's closer to right. Correct isn't going to be possible until Apple gets serious about color managing the thing right, but we can get better than the default, and that's all that this is attempting to do. Jun 24 10 07:11 am Link silverystars wrote: SRB Photo wrote: i'm just not sure that's going to work, technically speaking. from a color managed workflow standpoint, you don't convert images for device profiles, you convert them to a specific color space like sRGB into an image. the OS then uses the color space embedded in the image to map that to the device's profile. so, by trying to get a device profile for the iPhone screen and then converting an image to that profile, when the OS displays it, you're not going to get the result that you expect. color spaces like AdobeRGB and sRGB are device independent, that's why they're embedded into images so when the OS/device see it, it can say, "oh, i know how to transform the colors in this image with that color space to my display." Jun 24 10 07:59 am Link Thanks. I had the last iPhone profile for doing the same. I don't have the v4 yet but will keep this profile which will suffice for what I do. Jun 24 10 08:00 am Link silverystars wrote: I tested that. Jun 24 10 08:01 am Link silverystars wrote: The iPhone runs under a striped down version of OsX. The telephone is not color managed though at the level of a fully compliant ICC Colorsync workflow. Jun 24 10 08:05 am Link silverystars wrote: SRB Photo wrote: i would say that proves that it is color managed. the photo viewer on the iPhone looked at the embedded profile in each photo and displayed it properly for the device in hand. that's how ColorSync works. Jun 24 10 08:06 am Link silverystars wrote: silverystars wrote: i would say that proves that it is color managed. the photo viewer on the iPhone looked at the embedded profile in each photo and displayed it properly for the device in hand. that's how ColorSync works. *Sigh* Jun 24 10 08:15 am Link FWIW, I wouldn't put too much stock in the presently posted profile. As I give the unit more time of use, the screen is 'burning in' a bit with the white point shifting back towards 'right' and the gamma coming back under control. I'll reevaluate this weekend and if a profile is still needed update it then. This shift didn't occur with my 3G - I'm glad for it, but a warning would have been nice . Jun 24 10 09:02 am Link fascinating! I'll have to try spyder/ipad/vnc trick. You may be onto something! Jun 24 10 10:08 am Link I'm very excited to get the new iPhone. I have never had one and the new one is way too good to pass up. I already have a coming soon page on my web site for an iPhone 4 gallery of pictures. The camera is amazing! Jun 24 10 10:11 am Link This is awesome -- love that you're putting in the work on this. Thank you! Got mine today, too Jun 24 10 02:12 pm Link Updated the OP with a second version of the profile, and have done some more investigating. I was surprisingly (to me) close with my eyeball'd estimates for the color points in the original version; the new one is based on colorimetric measurement of the 'broken in' iPhone set at middle brightness. The problem with this, though, is that I have no way of adjusting the black and gray points (I'm creating this in PS), so until Datacolor get back to me about whether they can provide a workaround for lag while running a true calibration, this is the best that I can currently create. If anyone knows of good, free ICC editing software, that would also be an option, so please let me know. Jul 03 10 05:58 am Link New update - we're in business for a good profile. Perfect? No - auto brightness and its impact on color will always prevent that (or at least until Apple gets serious about color on mobile devices). As good as most of our monitors are calibrated? YES! So before I get wordy, here's the profile: iPhone4Final.icm. Simply "Convert to Profile" in PS, sync, and enjoy! Method (for those wanting to calibrate to your iPhone4, iPhone3gx, iPad, etc.): The key was finding a wonderful set of free calibration utilities - ones which aren't meant to go as fast as that included with my calibration HW. Namely, I'm now in love with the Argyll Color Management System suite of utilities (all command line!) as well as the makes-your-life-so-much-easier dispcalGUI, which provides a graphical front-end for things. The relevant settings in dispcalGUI are as follows: - Select your calibration device (be sure you install the ArgyllCMS USB drivers!) - Use the native whitepoint. - Use the maximal and minimal white and black points respectively - Set a gamma of 2.53 for the Tone Curve (measured previously - may be different on other devices!) - Black output offset at 100% - Black point correction at 0% - Rate at 400 - Calibration quality High - Profile quality High - Profile type "Gamma + matrix" - Testchart file "Default testchart for Gamma + matrix" profiles - Set the profile name as you like. On your iDevice, go into Settings and under General set "Auto-Lock" to never. Under Brightness, set Auto-Brightness to "Off". I created the calibration above with the slider at 50%. If you don't have one already, download and install a VNC server for your computer. Configure it to share just the center of the screen; don't let the client take control of the computer. Download and install a VNC client from the Application Store (I used Mocha VNC Lite). Connect your iDevice to the VNC server. Run the calibration as above - on my system it took ~40m. Save, convert, and enjoy! Jul 03 10 12:22 pm Link Wow. Jul 03 10 01:05 pm Link Graeme Gill is one of the best color scientist out there. Very helpful, very much into making open source Unix binaries , and contributing a lot for common users. It's beyond me command line, but for those who can the resulting profiles through Argyll, go far beyond any commercial profiler out there. Some of it is bleeding edge where if it goes past, you'll bleed. Yet a good profile for an iPhone is very useful to maximise it's rendering for those showing images on their phones. For this we thank you> Ps You might want to rename the internal name . I use Color Think , you probably can in ColorVison. Jul 03 10 01:18 pm Link To do an accurate colour matrix you would have to convert to unity gamma, matrix then convert back again. Since you iPhone will have an indeterminate gamma, probably around 0.45 but with tone curves, knees and slope limitation added to this it could be difficult. Matrixing the gamma corrected signal might give you accurate colours at one brightness level, but they won't track through the entire range. And simple matrixing won't work if the crossover points on your Bayer filter are not accurate, so more involved second order matrixing will be necessary. Jul 03 10 02:00 pm Link WMcK wrote: The calculations are handled by the software, thankfully. WMcK wrote: Measured at 2.53 (.395). WMcK wrote: It does (again, as closely as any calibrated monitor I've experienced), and the limitation for single brightness is acceptable, given that the out-of-box experience is far worse and that the brightness can be locked for critical (?) applications. WMcK wrote: What does a Bayer filter have to do with profiling an output device? Jul 03 10 02:08 pm Link SRB Photo wrote: WMcK wrote: The calculations are handled by the software, thankfully. WMcK wrote: Measured at 2.53 (.395). WMcK wrote: It does (again, as closely as any calibrated monitor I've experienced), and the limitation for single brightness is acceptable, given that the out-of-box experience is far worse and that the brightness can be locked for critical (?) applications. Sorry, my misunderstanding, I thought it was the camera on the iPhone you were profiling. When I read it more carefully I realised it was the screen, so most of what I said was rubbish, although there are some similarities in the profiling. Jul 03 10 02:44 pm Link Neil Snape wrote: Good call. I just asked my wife to do it in ColorSync . Done and uploaded. Thanks! WMcK wrote: No worries. I'm a sick / dedicated guy, but profiling the camera is well beyond my interests - at least for now . Jul 03 10 05:52 pm Link Another update : Datacolor have updated their Spyder3 software (the 4.0 version - I can't downgrade to test the 3.x series - please post if you've tried it) and it now works fine through the VNC method as above! Keys: - Tell your computer that you're profiling a projector. By popular request (*ahem*), there is now a 1 second delay between screen changes and the actual reading being taken, allowing full time for your VNC session to sync the window. - Use Native Brightness - I recommend using Native Color temp, but 6500K would be the alternate choice - I recommend using a gamma of 2.53; again, 2.2 will give good results as well - Save the profile to your system and convert iPhone portfolio images to that profile - Sync - Enjoy good color Jul 08 10 01:12 pm Link Great post, THANKS! Did a quick one for my iPad that might help others - http://www.kunar.at/download/iPadICM.zip Feedback welcome, Andi Aug 23 10 01:30 pm Link Missed this thread. Looks good. I'm bumping. Aug 25 10 12:56 pm Link |