[interested_observer]

Ok, after years of fighting it - I finally purchased Adobe's LightRoom/PhotoShop package for a number of reasons. I've had LR 5.5 for several years now - but I need LR/PS CC for a number of reasons. I've read a lot on the web the last couple of days, but

• How does one go about processing Pixel Shift images in LR/PS?

• When you view a PS image in LR are you viewing the in camera processed raw image, or does LR un-bundle the 4 embedded images, process them into a single new resultant image?

• When you use PS and bring in a PS image, is pulls up ACR and then does what with the 4 embedded images?

I've found absolutely no information on the web, but a couple of threads here - but no real information on these items.

At least in RawTharapee, when you load a PS image, it identifies it so that you can select the PS DeMosaicing method - and you have a good idea of what you are getting.

[pschlute]

How does one go about processing Pixel Shift images in LR/PS?

When you view a PS image in LR are you viewing the in camera processed raw image, or does LR un-bundle the 4 embedded images, process them into a single new resultant image?

When you use PS and bring in a PS image, is pulls up ACR and then does what with the 4 embedded images?

In both programs treat the PS raw file as if it was any other raw file.

In LR you are viewing the un-processed raw file. Simply make your raw adjustments and save. LR does not take the 4 images apart.

In PS you are working in Adobe Camera Raw. Same as above, just process as normal and save as your required image format eg. jpeg.

*Unprocessed except for any default Adobe (or other) profiles you may have selected when opening a new file.

[rechmbrs]

In both programs treat the PS raw file as if it was any other raw file.

In LR you are viewing the un-processed raw file. Simply make your raw adjustments and save. LR does not take the 4 images apart.

In PS you are working in Adobe Camera Raw. Same as above, just process as normal and save as your required image format eg. jpeg.

*Unprocessed except for any default Adobe (or other) profiles you may have selected when opening a new file.

PS and LR both perform the compositing of the four Pixel Shift shots which are contained in the DNG or PEF and then outputs a single file. Neither have a movement compensation ability. Even though it is more cumbersome to use, RawTherapee is my preferred method of handling Pixel Shift files. In that Sony is offering Pixel Shift, Adobe recently has showed some interest in the movement problem but I haven't heard anything about updates to ACR or LR.

RONC

[interested_observer]

Thanks for the responses, they helped a lot.....

In both programs treat the PS raw file as if it was any other raw file.

In LR you are viewing the un-processed raw file. Simply make your raw adjustments and save. LR does not take the 4 images apart.

In PS you are working in Adobe Camera Raw. Same as above, just process as normal and save as your required image format eg. jpeg.

*Unprocessed except for any default Adobe (or other) profiles you may have selected when opening a new file.

PS and LR both perform the compositing of the four Pixel Shift shots which are contained in the DNG or PEF and then outputs a single file. Neither have a movement compensation ability. Even though it is more cumbersome to use, RawTherapee is my preferred method of handling Pixel Shift files. In that Sony is offering Pixel Shift, Adobe recently has showed some interest in the movement problem but I haven't heard anything about updates to ACR or LR.

RONC

So, what I have found through a reasonable amount of sleuthing and experimentation is the following....

• Both LR and ACR - will upon initial loading of the PS image, will DeMosaicing the 4 individual packed PS images, and re-render the image, and not use the in-camera created resultant image.

• Raw Tharapee - performs the same function, given that for the DeMosaicing method - Pixel Shift is specified, and you specify the number of images to use (1,2,3 or 4).

How did I determine this. Well, I had taken a 4 shot pano and assembled the pano several ways 1) using the LR pano stitch function and 2) using the Microsoft ICE utility. They are ever so slightly different. How so? The LR pano had some artifacts (used the LR/ACR re rendered images), while the ICE pano had none of the artifacts (used the in-camera produced image that is displayed on top of the 4 image packed file).

I shot 4 frames for a pano right at sunset, and with the light changing fairly rapidly, at 4x (yes pixel peeping) and beyond, there are movement artifacts - dynamic changing light is "assumed" by the DeMosaicing algorithm to be physical movement. This is shown in the following link, where light trails are taken as movement, creating artifacts - the same type of artifacts I'm getting, but substantially less visible. I'm seeing the artifacts on solid rock faces that are not moving (even with the California earthquakes).

• What happens if you try to pixel shift a long exposure with movement. ? David Abbs Photography

So, that puts the questions I had to bed for me. The additional detail captured in the PS images is absolutely striking.

.... I added two example images

• In the first file, which is DeMosaiced and stitched using LR, you can see in the sky some horizontal lined artifacts.

• In the second file, which is via ICE just uses the in-camera produced image, which are stitched does not have any of the artifacts, and lacks a lot of the detail.

• And the third one is the entire 4 frame LR pano, just for reference.

Attached Images

View Picture EXIF

File Date

     

---

Last edited by interested_observer; 07-09-2019 at 08:46 PM.

[Rondec]

Thanks for the responses, they helped a lot.....




So, what I have found through a reasonable amount of sleuthing and experimentation is the following....

• Both LR and ACR - will upon initial loading of the PS image, will DeMosaicing the 4 individual packed PS images, and re-render the image, and not use the in-camera created resultant image.

• Raw Tharapee - performs the same function, given that for the DeMosaicing method - Pixel Shift is specified, and you specify the number of images to use (1,2,3 or 4).

How did I determine this. Well, I had taken a 4 shot pano and assembled the pano several ways 1) using the LR pano stitch function and 2) using the Microsoft ICE utility. They are ever so slightly different. How so? The LR pano had some artifacts (used the LR/ACR re rendered images), while the ICE pano had none of the artifacts (used the in-camera produced image that is displayed on top of the 4 image packed file).

I shot 4 frames for a pano right at sunset, and with the light changing fairly rapidly, at 4x (yes pixel peeping) and beyond, there are movement artifacts - dynamic changing light is "assumed" by the DeMosaicing algorithm to be physical movement. This is shown in the following link, where light trails are taken as movement, creating artifacts - the same type of artifacts I'm getting, but substantially less visible. I'm seeing the artifacts on solid rock faces that are not moving (even with the California earthquakes).

• What happens if you try to pixel shift a long exposure with movement. ? David Abbs Photography

So, that puts the questions I had to bed for me. The additional detail captured in the PS images is absolutely striking.

.... I added two example images

• In the first file, which is DeMosaiced and stitched using LR, you can see in the sky some horizontal lined artifacts.

• In the second file, which is via ICE just uses the in-camera produced image, which are stitched does not have any of the artifacts, and lacks a lot of the detail.

• And the third one is the entire 4 frame LR pano, just for reference.

Just to be clear, when you are choosing a number in Raw Therapee, it is not how many images to use in a pixel shift image, it is which of the images should be used as the "base" image if there is some movement between frames. If there is no movement, there will be no difference between Lightroom and RT. If there is movement, RT can handle it in a variety of ways.

[interested_observer]

Just to be clear, when you are choosing a number in Raw Therapee, it is not how many images to use in a pixel shift image, it is which of the images should be used as the "base" image if there is some movement between frames. If there is no movement, there will be no difference between Lightroom and RT. If there is movement, RT can handle it in a variety of ways.

Thanks for that information!!!! I didn't realize that. I've been chasing a number of items - using the most recent take from this road trip we took on July 3 - a 14 hour goat rope that turned out to be almost epic.

One more question, and I've contemplated this for quite a while now. Unfortunately I didn't have the presence of mind when I was out shooting to give it a try. I had the perfect opportunity, but didn't think to follow through.

When I am shooting the foreground landscape for an AstroTracked image (Milky Way over some landscape) - I usually shoot at least a 2 minute landscape frame (it's dark out). This time it was soooo dark, I shot a series of 4 minute frames (and now I feel I should have shot for 6 minutes). I'm really seriously considering shooting sets of 30 second PS frames (2 minutes of "light" in total for each PS frame), perhaps 1, 2 or 3 sets, and then stack them together. My thinking is I can either shoot a 2, 4 or 6 minute frame, or 1, 2 or 3 PS frames - same amount of light, however the PS frames are essentially "post processed" to a degree, providing the potential of additional color, lower noise, and greater dynamic range - all good. It essentially takes no additional time to do this, and there does appear to be some additional potential benefits. It certainly can't do any harm (famous last words). I'm kicking myself for not realizing this at the time. Also, since with the PS frames, the 2 minute shots are broken into 4 - 30sec shots, and the noise - especially the white dot noise, should be averaged down if not out, along with the effective ISO value averaged down.

One downside to this idea, is - there is this location I want to start shooting in a few weeks, that has a set of road switchbacks down the side of a steep canyon wall. The PS approach will not work well at all with this location due to the changing headlights (and physical motion). So, I think that this potential approach needs to be judiciously applied based on the situation at hand.

We were out in a Bortle 1 location, no moon, clear skies, no breeze - absolute darkness. The sky was lite up with the stars so much, I cast a shadow on the ground, with just the ambient starlight.

[rechmbrs]

Pixel Shift is not a stacking process and neither RT nor LR de-mosaic the PS data as removal of the Bayer pattern is inherit in the process of compositing the four shots. When using RT in handling the movement compensation, RT applies a de-mosaic process and uses that data in areas deemed to be needing movement compensation and does compositing over the rest of the image.

RONC