Pages: [1]   Go Down

Author Topic: Super Resolution  (Read 1952 times)

PeterAit

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4559
    • Peter Aitken Photographs
Super Resolution
« on: March 16, 2021, 11:25:38 am »

I have been playing with the Super Resolution tool in the new ACR. Very cool! In a nutshell, it increases the resolution of a RAW file by 4x. It seems to do a really good job and to be a "free" alternative to Gigapixel.
Logged

Rajan Parrikar

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3950
    • Rajan Parrikar
Re: Super Resolution
« Reply #1 on: March 16, 2021, 03:25:12 pm »

I have been playing with the Super Resolution tool in the new ACR. Very cool! In a nutshell, it increases the resolution of a RAW file by 4x. It seems to do a really good job and to be a "free" alternative to Gigapixel.

This feature may well have killed Topaz Gigapixel.

I tried ACR Enhance on a photo of peacock plumage. It scales both the dimensions 2x. Then I compared it to Gigapixel (with zero Blur reduction and zero Noise Suppression). ACR looks more natural when viewed at 400%, without the edge artifacts seen with Gigapixel.

And ACR takes just 3 seconds whereas Gigapixel took well over 2 mins.


digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20648
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Super Resolution
« Reply #2 on: March 16, 2021, 03:31:59 pm »

In addition, you've got actual raw data:
https://blog.adobe.com/en/publish/2021/03/10/from-the-acr-team-super-resolution.html#gs.vswudd
Quote
Your computer will put on its thinking cap, crunch a lot of numbers, then produce a new raw file in the Digital Negative (DNG) format that contains the enhanced photo. Any adjustments you made to the source photo will automatically be carried over to the enhanced DNG. You can edit the enhanced DNG just like any other photo, applying your favorite adjustments or presets. Speaking of editing, I recommend taking another look at your Sharpening, Noise Reduction, and possibly Texture settings. All of these controls affect fine details, and you may need to tune these for best results on the enhanced photo.
Not the case with Gigapixel is it?
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

NikoJorj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1082
    • http://nikojorj.free.fr/
Re: Super Resolution
« Reply #3 on: March 17, 2021, 06:18:30 am »

Isn't that a linear DNG, rather than a "true" (Bayer/Xtrans) raw?
From what I infer from Eric Chan's blog post, Enhance details is applied first, and that applies demosaicing, doesn't it?

And I've seen an example where Super Resolution worked quite poorly, it was a focus stacked macro TIFF from a Leica Monochrom (whereas the results were quite good with a single raw from that camera). See this forum post in french (100% crops, Adobe SR at the left, Gigapixel at the right - more details, always in french, at the bottom of the previous page).
Seems the machine learning was self-isolating at the time of the 'shells' lesson?

BTW, there already was a discussion about that at https://forum.luminous-landscape.com/index.php?topic=137874.0 (but the title was perhaps a tad cryptic).
Logged
Nicolas from Grenoble
A small gallery

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20648
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Super Resolution
« Reply #4 on: March 17, 2021, 11:08:34 am »

Isn't that a linear DNG, rather than a "true" (Bayer/Xtrans) raw?
I posted that question yesterday in the Adobe pre-release forums for Eric or other engineers and await an answer.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Chris Kern

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2034
    • Chris Kern's Eponymous Website
Re: Super Resolution
« Reply #5 on: March 17, 2021, 08:50:27 pm »

Isn't that a linear DNG, rather than a "true" (Bayer/Xtrans) raw?
From what I infer from Eric Chan's blog post, Enhance details is applied first, and that applies demosaicing, doesn't it?

Yes, the output files appear to be linear DNGs.  When you compare the file sizes of the raw sensor data with those enlarged by ACR Super Resolution, it's clear the latter have been demosaiced:

-rw-r--r--  1 xx  staff    45M Jan 12 13:21 DSC_5178.NEF
-rw-r--r--  1 xx  staff   531M Mar 10 15:35 DSC_5178-Enhanced.dng

-rw-r--r--  1 xx  staff    26M Nov  9 15:17 XTTH0039.RAF
-rw-r--r--  1 xx  staff   320M Mar 10 20:21 XTTH0039-Enhanced.dng

Still, they're "raw enough" for most practical post-processing purposes.
« Last Edit: March 17, 2021, 11:30:14 pm by Chris Kern »
Logged

Chris Kern

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2034
    • Chris Kern's Eponymous Website
Re: Super Resolution
« Reply #6 on: March 17, 2021, 08:53:25 pm »

BTW, there already was a discussion about that at https://forum.luminous-landscape.com/index.php?topic=137874.0 (but the title was perhaps a tad cryptic).

Yes, sorry about the subject line: I'm a frustrated headline writer.  That post does include some sample image files.

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com
Re: Super Resolution
« Reply #7 on: March 23, 2021, 12:25:54 pm »

it's clear the latter have been demosaiced

The opposite would be very strange, difficult to achieve and in the end would provide no advantage. The good thing of RAW files is that they are linear so that all colour and shadows/highlight information is there ready to be enjoyed. The Bayer structure is not an advantage per se, rather a disadvantage.

Regards

SmartSolutions

  • Newbie
  • *
  • Offline Offline
  • Posts: 2
    • Smart Solutions Taxation & Accounting
Re: Super Resolution
« Reply #8 on: March 25, 2021, 04:22:44 am »

Thanks for sharing this topic
Pages: [1]   Go Up