If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
Comparison 50MP HHHR vs Photoshop Super Resolution
Photoshop ACR now allows to increase the resolution by a factor
of 4. You open a 20MP RAW file and get an 80 MP image. I've compared it to the 50 MP handheld hires mode of the Olympus E-M1 III (the camera takes sixteen 20MP images and merges them in-camera into a 50MP image; no tripod is needed). I took this image in 50 MP handheld hires mode, getting two RAW files (a 50MP and a 20 MP one). Here is the original image, resized to 1024x768: https://myolympus.org/images/comp_hi.../P8220381_.jpg Here is a comparison of two crops of the 50MP (handheld hires, left side) and the 80MP (Photoshop Super Res, right side). First crop (top part, the clock): https://myolympus.org/images/comp_hires_acr/detail0.jpg Second crop (bottom part, station entrance): https://myolympus.org/images/comp_hires_acr/detail1.jpg You can see that the 50MP handheld hires has real detail, while the 80MP Photoshop Super Res has a lot of fake detail and noise. For instance the net left and right of the clock and the brick work in the first crop are rendered much better in the 50MP HHHR image. In the second crop the small text in the red boards is readable in the 50MP image, while it's not in the 80MP image. However there are moving people in the image, which result in ghosting in the 50MP HHHR image. One option would be to upres the 20MP image with Photoshop Super Res to 50MP and then mask away these image areas with ghosting. -- Alfred Molon Olympus 4/3 and micro 4/3 cameras forum at https://groups.io/g/myolympus https://myolympus.org/ photo sharing site |
#2
|
|||
|
|||
Comparison 50MP HHHR vs Photoshop Super Resolution
In article a218700b-f09b-4c84-b0f1-
, Magani says... Be interesting to see an A/B comparison between this and Topaz Gigapixel AI. Here is a comparison between Olympus pixel shift and Gigapixel AI: https://www.youtube.com/watch?v=ogQXLIHFmTA&t=750s And here Gigapixel AI looks a lot better than Photoshop Super Resolution. However also Gigapixel AI generates fake detail. I can't find the link anymore, but I remember another comparison where if you looked in image areas with random detail (for instance without geometric patterns), you could see that Gigapixel AI was just generating fake detail. -- Alfred Molon Olympus 4/3 and micro 4/3 cameras forum at https://groups.io/g/myolympus https://myolympus.org/ photo sharing site |
#3
|
|||
|
|||
Comparison 50MP HHHR vs Photoshop Super Resolution
In article 53d37fcd-fd9d-48fd-be36-
, RichA says... They all generate fake detail. How can they not? Well, they could identify the scene/subject and add the right pattern from a library. But most likely these algorithms are not that smart yet. -- Alfred Molon Olympus 4/3 and micro 4/3 cameras forum at https://groups.io/g/myolympus https://myolympus.org/ photo sharing site |
#4
|
|||
|
|||
Comparison 50MP HHHR vs Photoshop Super Resolution
In article ,
RichA wrote: They all generate fake detail. How can they not? Well, they could identify the scene/subject and add the right pattern from a library. But most likely these algorithms are not that smart yet. Then it's still fake because the camera didn't actually see it. Can't really attribute any more reality to the scene than if it was generate artificially. all photos are faked to some extent, some more than others. |
#5
|
|||
|
|||
Comparison 50MP HHHR vs Photoshop Super Resolution
On 14/03/2021 7:06 pm, RichA wrote:
On Sunday, 14 March 2021 at 08:29:18 UTC-4, Alfred Molon wrote: In article a218700b-f09b-4c84-b0f1- , Magani says... Be interesting to see an A/B comparison between this and Topaz Gigapixel AI. Here is a comparison between Olympus pixel shift and Gigapixel AI: https://www.youtube.com/watch?v=ogQXLIHFmTA&t=750s And here Gigapixel AI looks a lot better than Photoshop Super Resolution. However also Gigapixel AI generates fake detail. I can't find the link anymore, but I remember another comparison where if you looked in image areas with random detail (for instance without geometric patterns), you could see that Gigapixel AI was just generating fake detail. -- Alfred Molon Olympus 4/3 and micro 4/3 cameras forum at https://groups.io/g/myolympus https://myolympus.org/ photo sharing site They all generate fake detail. How can they not? The image only contains as much detail as it does. There may be "hints" of greater detail but even "A.I." can only extrapolate so much. It's like those ridiculous scenes on police shows where they "zoom in" on a crappy video image, then say, "can you sharpen it up?" and voila, detail magically increases 10-fold. I've been using "Pixbim ColorSurprise AI" to convert some old family B&W photos to color and sometimes it's amazing at the apparent increase in detail. Unbelievable at times! |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Sigma announces APS-H sensored cameras, claim 50mp resolution | Alfred Molon[_4_] | Digital Photography | 3 | February 26th 16 05:11 PM |
Sigma announces APS-H sensored cameras, claim 50mp resolution | nospam | Digital Photography | 0 | February 25th 16 05:37 AM |
When is 24-50mp not 24-50mp? When the lens won't support it | Sandman | Digital Photography | 9 | December 22nd 15 05:11 PM |
When is 24-50mp not 24-50mp? When the lens won't support it | Eric Stevens | Digital Photography | 1 | December 20th 15 03:33 PM |
Elements vs Photoshop - comparison of features/abilities question | Steven Wandy | Digital Photography | 8 | October 21st 08 10:13 PM |