Superresolution Photos with Photoshop

I do a ton of photomerging images but, I've never used this newly found technique...
https://www.youtube.co​m …ion+Photos+with​+Photoshop

Now I gotta admit that I don't understand how it works, if it does. Can anyone help me out?

C'mon folks, someone's gotta understand it.

Maybe a link to something specific would help, rather than a suggested search in Youtube. Oh and personally I would prefer the info to be discussed to be written words, not a video. IIRC PS has a pixel limit, but it is pretty big, I think that there is a file size limit for PSD files, not sure what that is possibly 3GB? You can get round that too though I believe by using a related file format, PSB seems to ring a bell.

Alan

I watched the video put out by Adorama, Mark Wallace. If you watch it, he fully explains this method. One thing you can count on, a single image might take hours to produce. The file sizes are huge.

https://www.youtube.co​m/watch?v=0tGR-Q9Pkjc

https://www.youtube.co​m/watch?v=o7POpid-e8U
there are several that illustrate the same thing...by shifting your perspective a few pixels and merging them,
you somehow achieve that super resolution. That's where I get lost.

I think this is a nice simple explanation:

http://petapixel.com …on-photos-with-photoshop/

The process works to the extent it does because some or all of the almost identical images have some good data that's missing in others. Keep the good, ignore the bad, and gain some apparent resolution.

I remember using a software called Photoacute Studio a long while ago to do this with my 5D2, I even got a free license by contributing the 85LII profile. Problem is that it requires you to profile your camera+lens combo to work, but it was pretty nifty for studio work.
Now that I'm using a 645Z, it's of course no longer needed, but I'm interested to see that it can also be done in Photoshop to some extent, those few years back smart layers and the align tool didn't exist yet.

The super resolution technique, as detailed in the PetaPixel article, is essentially similar to the kind of techniques used in astrophotography - you are averaging multiple, identically shot images of a scene to increase the signal-to-noise ratio on a per-pixel basis. Because the resulting pixel value in an image is an interpolated result of the surrounding sensor values, the more data you feed to the process, the better the "signal" (the "actual" pixel value, compared to noise and other artifacts) will be.

If you shoot a static scene and slightly change the position of the light falling on a particular pixel with each successive image, then you can combine the slightly shifted images to get a better representation of the signal in the neighborhood of that location on the sensor. Upres-ing the source images permits the combination to eliminate spatial issues like aliasing and moire, as the combination process will smooth these artifacts as more data are fed to the combining algorithm.

The more data you have about a scene, the more "true" signal you get at each pixel.

The Adorama video linked above is not about super resolution - it is about combining several narrow field-of-view images to create a wider field of view image with more pixels than the source camera's sensor has. This is just pano stitching. The resolution of the image (the discrimination of detail, noise, etc.) is identical to all of the individual source images. There is no spatial or temporal averaging or combination going on that will change the signal-to-noise ratio of the final image, you will just have more pixels in the final image than in a single, equivalent field of view image shot with a wide angle lens.

kirk

This Photoshop-based technique might be beneficial to folks who shoot images using the Magic Lantern DualISO utility - the resulting images suffer from moire and other resolution-based artifacts (namely aliasing and moire) due to the alternating pattern of the two ISO images being read from the single sensor. One could probably achieve pretty good results with not a lot of images. I will give it a shot and see if it makes a difference.

kirk

I posted a similar query on LuLa which lead to this thread...http://forum.luminous-landscape.com/index.ph​p?topic=113386.0
proved interesting.

too much work on something you just wont see on average size prints.

Good for giant size prints with a magnifying glass assistance.

I just get the impression that one would get the high resolution that the high MP beasts obtain.

Talley wrote in post #18126573
too much work on something you just wont see on average size prints.

Good for giant size prints with a magnifying glass assistance.

While it is true that the resolution difference may not be noticed in average sized prints, the blending of the layers (which I would do with a Smart Object and median blending mode, not by altering layer opacities to different amounts) is a great way to reduce noise. This was evident in Ian Norman's video, referenced above. I do this with Milky Way shots with only eight exposures at ISO 3200 or 6400, and there is very little noise left when I am finished.

chauncey wrote in post #18126692
I just get the impression that one would get the high resolution that the high MP beasts obtain.

That is exactly what you get, only higher resolution in most cases. There may be drawbacks (aside from the labor), but the results look impressive.

As Kirk mentioned, in astrophotography this a key process called dithering and also helps a lot with noise removal. This process can be really useful for taking low noise, high details photos in low light condition.