Are you doing the pano stiching after you processed the photos in ACR or Lightroom? If that is the case, you can apply the profile based lens correction before pano merge in Photoshop. You'll get better result this way. There is also a pano merge setting to compensate for vignetting in PS. That will also help further reduce the vignette effect for pano merge.
As to your question of getting lens profile for your phone camera, I don't think Adobe created a profile for it. As long as the camera embeds the necessary EXIF metadata in the photo it took (focal length, f-number, optionally focus distance), you can use the Adobe Lens Profile Creator to create a lens profile for your phone camera.
Your reply doesn't answer any of my questions.
How to create a profile if the camera doesn't support an exposure lock?
I would guess you don’t absolutely need an exposure lock as long as the camera records the exposure parameters within each photo, so the LPC can compute a consistent light level. It would help to have a consistently lit and relatively neutral background, such as a blank wall behind the calibration target.
Unfortunately, a guess is not helpful in this case. Did anybody try it out? The camera *does* report the exposure parameters (even very exactly so and only shutter speed varies). If LPC uses EXIF to compute a consistent light level, wouldn't it be mentioned in the user guide?
By trying it out, I mean a comparison of vignetting profiles created with and without a locked exposure setting.
I am guessing because there are lens profiles for cameras like an iPhone that presumably don’t have an exposure lock, although I suppose it is possible that some sort of camera-app for an iPhone does have M mode.
The math seems easy enough to me to compensate for differing exposures and the people that figured out how to do geometric lens corrections know a lot more about math than I do.
I am sorry but guessing does not solve my problem.
The iPhone could have a bad profile or be provided by Adobe or whatever. Moreover, the math is trivial if the camera's tone curve is known. And the latter isn't in general. LCP could calibrate for a camera's unknown tone curve but following the user guide's instructions, it lacks all the required meeasurements to do so.
As I pointed out, the camera settings that could affect vignette amount are the focal length, f-number and focus distance, in decreasing order of influence. The major influencer is the focal length. The other two are minor factors. Unless you are shooting macro, you can pretty much ignore the f-number and focus distance factors in building your own lens profiles. Since you care most about shooting pano, you can save yourself from building a more comphrehensive lens profiles by just considering varying the focal length when building the lens profiles.
If you still want to build a comphresensive lens profiles with all focal length, f-number and focus distance variants, since you cannot control the f-number, you need to control the ambient lighting in your shooting environment.
simonsaith, I fear but you have not understood my problem. I know everything you said. I don't care about a complete profile, just the one for the f-number and focal length (both are constant anyway) and distance (infinity) I am using.
Nevertheless and you may lack this knowledge, LCP requires to shoot multiple test shots (typically 9 or 15) with a single given set of parameters. And the problem is what to do if the camera cannot keep a constant exposure between those 9 shots (as is a requirement according to the LCP user guide).
I see that you have a "staff" icon. So, how about forwarding the problem to the engineers who wrote LCP? I fear but its only them knowing the answer.
Personal attack is not helpful to resolve your issues. Maybe you did not ask your specific question clearly so that I can understand. Just FYI that I am the Adobe staff who help created Adobe Lens Profile Creator and the rest of the lens correction eco-system now in place.
Now that I understood your question, you have a few options, some of which I have mentioned earlier:
1) Control your ambient lighting environment when you shoot the checkerboard. The goal is try to induce your camera to shoot with constant exposure for the multiple shots.
2) If you cannot succeed on this, Photoshop's "Auto Align Layers..." command has a "Vignette Removal" option to do pano merge while automatically compensate for the vignette.
3) Try the Lightroom flatfield plug-in http://labs.adobe.com/technologies/lightroomplugins/, which uses a different mechanism (single shot) to correct the vignette among among other things. If you cannot keep the f-number/focus distance constant when switching between the calibration shot and real shot, you might just need to assume that differences caused by the f-number/focus distance are small enough to be negligible.
Hope this is helpful,
Ok, I see the answer is to vary the ambient illumination to get a consistent exposure value via trial-and-error.
One way I see this working by taking a series of shots with the target at each position, and seeing which ones used a different exposure, computing the EV difference of those, then vary the illumination using a light meter (or DSLR in M-mode) as a way to measure how much brighter or dimmer to make the illumination.
As far as how to make the illumination brighter or dimmer, without studio lighting, a dimmable incandescent behind a diffusing sheet of something translucent, or if you don’t have something dimmable, put it behind a large diffuser and shade it with various objects in front, though a dimmer switch would be the easiest.
One thing not to do would be use fluorescent lighting or similar light source which varies in intensity and color 100 or 120 times a second depending on the local’s A/C frequency so with a fast shutter speed can be much dimmer or brighter.
@simon, sorry you took it as a personal attack. But you didn't answer my question, no idea why you took that personal. Anyway.
Now looking at your answer I am sorry to say that this wouldn't work (maybe except #3). In detail:
Ad 1) This cannot work. There is a requirement for:
- constant exposure setting among shots for a single given EXIF set of parameters (my problem because shutter would vary)
- constant ambient lighting! (Of course, as otherwise, it would be impossible to calibrate for change of brightness between center and corners; it's described in the user guide too)
Ad 2) I do not use PS to do pano stitching. The result can't compete with Autopano. PTGui has the same option and the same problem. Autopano Giga actually can read Adobe profile files.
Ad 3) This may actually work with an even enough background. Thanks for the hint. It is one of several options I can go if LPC cannot be used.
So, Simon, am I right to read your answer to my question (this thread's fisrt line: "How do I create an ACR custom lens profile if camera exposure can't be set to manual?") as follows: You can't. Correct?
Maybe, you actually forward the question to one of the developers. My question still stands. Esp. if varying exposure information in the EXIFs is somehow corrected for within APC.
@sprengel, thanks for all your help to figure out how the ambient illumination may be controlled. However, as I pointed out above, it can't work. LPC needs "to see" that the target appears darker when in the corners. Of course, as this is what the vignetting profile is created from. LPC does not need an even illumination and this is the price to pay (it must remain constant).
You are correct in saying that the brightness of the image changes when the camera metering changing the exposure (in your case the shutter) depending on what area of the field-of-view the target covers—this would most likely occur if the camera was set to center-weighted rather than average metering, so varying the ambient light would merely be a way of getting the same EXIF into the photos; however, because phone-cameras can be profiled, it’s possible the LPC can compensate for this, perhaps by normalizing the brightness across the edges of the charts between adjacent shots. I agree it would be nice to hear from someone technical at Adobe to say Yes the LPC can compensate for changes in overall brightness or No it cannot.
Simplier than varying the ambient lighting would be to use a metadata rewriter tool (maybe EXIFtool) to change the shutter speed in your jpgs if that is one of the parameters that can be overwritten.
While waiting, perhaps indefinitely, for Adobe to answer your question, can you just try making a lens-profile and see if it corrects vignetting reasonably well or not?
I also use AutoPanoGiga.
>Yes the LPC can compensate for changes in overall brightness or No it cannot.
No it cannot. If you cannot induce your camera to keep the exposure constant between the shots, ALPC cannot guanrantee the correct results. EXIF metadata can always be tweaked using some 3rd party tool like ExifTool.
Really, you should talk to your camera vendor to update the camera software to allow you to turn off the auto exposure, and preferrably also allow you to manually set the f-number.
If you read the first post, this is the camera-app of a 41MP Nokia Pureview 808 camera-phone, so maybe someone could write a new camera app but the phone is Symbian OS which is dead, making that unlikely.
It would probably be easier for Adobe to rewrite the LPC to allow vignetting to be computed from a single shot of a blank wall without a target in the shot, than to have someone rewrite a camera app that allows manual exposure.
Another idea about how to get the phone to keep a constant exposure would be to experiment with putting darker and lighter objects in the field-of-view away from the target area to make the camera metering adjust things so it’s exposure is the same from one shot to the next. This would take some doing but should work, unless the LPC uses the part of the frame that doesn’t contain the checkboard target in its computations.
You’re basically varying the scene around the target so the camera takes the same exposure of the target each time.