RedTail Hawk; comparison with Adobe super resolution, no enhancement, and gigapixel AI

this image is enhanced with the super resolution offering in camera raw.

same image no enhancement

same image with Topaz gigapixel AI and Topaz De noise

Here is another set of images using the super enhancement tool in camera raw.
The first image posted is enhanced with pixel count doubled by the super enhancement tool.
The second image is unenhanced and straight from the original image.
The third image is the original unenhanced image with normal pixel count, APC mode or roughly 20 megapixels for the full image.

Although there might be some exposure variations between the two enlarged images, please look at the resolution for your comparison. I have tried to do the same amount of cloning and exposure adjustment with both images but that is not perfect.

This tool is accessible either through camera raw via Photoshop or in Light Room by right clicking on the image. This is a raw image. I have been told one can also use JPEG’s but I have not tried that. It does not work with a tiff image via light room or Photoshop.

Specific Feedback Requested

The enhanced image is roughly 2350 by 1730. First image size
The unenhanced image is roughly 1120 x 780. Second image size. I forgot to lighten the shaded part of the face on the second image.
Un cropped APC mode is 5620 by 3745. This is the size of the third image.
Full frame on this camera is 8640 x 5760.

All three images are processed in Light Room, Photoshop, and Topaz De Noise. No other processing.

Technical Details

Is this a composite: No
Iso-640, 400+1.4 X, F5, 2000th

That’s a heck of a crop! The second image shows improved resolution & details and also seems to have eliminated some chromatic aberration which is apparent in the first image.

I don’t think the same technology exists in Lr yet although it is coming. The Enhance Details right click thing is a few years old so far as I know.

In these images I can see a very significant difference, David. The plumage is much crisper in the first image and it shows even more in the fir needles which look quite sharp in the first image and rather blurred in the second. I tried this on an image last night, but it may not have been a good choice because the improvement didn’t pop out at me. It also wasn’t a huge crop which might make a lot of difference.

I suspect the size of the crop makes a difference. One can also access this tool and camera raw through Adobe Bridge. Online reviews state it works with TIFF images though I have had some problems doing that. Files are saved as .DNG. Other online sources state it is good for improving older low resolution images. I have yet to try that.

The first image does look sharper to me. But, is the contrast difference the result of the enhancement? Also, when I’ve tried this the pixels stay the same dimension wise, but the file size is huge. How are you getting the larger pixel dimensions? In any case, the detail with this much cropping is impressive-in both images. You have a nice setting for the hawk and the background fits in well.

Interesting test. One that I see missing is opening the image in Photoshop, then using Preserve Details 2.0 (in Image/Image Size) as the interpolation method to double the image size as the Super Resolution feature does. I’ve tried it with about a dozen images. I tried images with fine detail, images of distance landscapes with much broader detail, people images, etc. So I created an image with Super Resolution and one from the original RAW interpolated to 200% using Preserve Details 2.0. Rather than letting my eyes try to see the difference I put them in a PS document as layers and set the top layers blending mode to Difference. Then PS will look at the images. Any pixel that is different will be something other than black. In all of the images I tested if you weren’t looking at 100% you couldn’t see anything. At 100% you could see a gray outline of the finer details. Applying a small amount of sharpening to the Preserve Details 2.0 image made many of them nearly identical.

I think this technology has some promise over time. This is iteration number one and I’m guessing it will continue to improve. The workflow now is really clumsy as you have to open the image directly in ACR. It isn’t in LR yet, but is promised to be coming so that will help with the workflow problem except it makes a copy as a DNG.

I’ve seen several YouTubers exclaiming it is a miracle as they zoom into 200% on their monitors and try to tell their viewers how much better it is, which just makes me laugh. Anything zoomed beyond 100% is meaningless to discern detail as it is beyond the resolution the monitor can produce.

Thanks for the comment Keith. I’ve never really used the Preserve Details 2.0 selection and I did give it a try. Very interesting and I agree the technology is starting to get there. As mentioned previously, Topaz gigapixel AI is a similar product but I have not had the same results as with the super resolution feature.
With respect to Light Room, there is an option to right-click on your image and use enhance but it doesn’t double or quadruple the pixel size as does the super resolution option. I’m not sure what the differences between that and the PRESERVE DETAILS 2.0.

Hi David: The option in Lightroom for “Enhance Details” has been there for several years. It was introduced primarily for the Xtrans sensor that Fuji Cameras use. It just tries to do some fancy sharpening. As you noted, it does nothing in terms of interpolating to a larger pixel count. Adobe has said that the new “Super Resolution” feature will be coming to LR. I’m surprised it didn’t come out at the same time since LR uses the same RAW engine under the covers as Adobe Camera Raw. Different user interfaces, but the same underlying code base. They just need to add the new feature to the UI in LR.

That is an algorithm to do interpolation for increasing the number of pixels in an image. It is very good and is the best option for increasing the size of an image in PS. My personal take on third party plugins like Gigapixel is that they are not needed for 99%+ of the work that most of us do. If you planning on making an image that is multiple meters wide then that might be a reason. For everyday ordinary size increases, I don’t see any benefit.

Since I have gigapixel AI, I went ahead and applied it to the image before I did any cloning or noise reduction. Gigapixel gives one the option to double, quadruple or increase the pixel count by six times. It is very slow compared to any of the Adobe offerings. I also find that there was an increase in noise that was not removed by Topaz Denoise AI. I went ahead and posted the gigapixel image which is the third image from the top in the original posting.

I agree that there are no benefits to using gigapixel for this.

I just looked at the Gigapixel results you posted. Looking at it I’d say that not only are there no benefits, but it botched the image enlargement adding all kinds of artifacts with no improvement in details.

Absolutely. The best enhancement is the first image. Topaz needs to coordinate the Gigapixel enhancement with DeNoise AI. Their products are pretty good but sometimes I think they could do a better job with integrating their software offerings. For me, Topaz De noise AI is one of the best things they have. Most of the time I use DxO photo lab 4.2 to process the raw image and then apply Topaz de noise AI. That seems to be the best workflow for these high pixel count cameras.
Light room just added the ability to open files from my newer camera but DxO has not done that yet so I am not able to apply that process currently. It typically takes DxO several months to accommodate newer cameras.

Hi David
My vote is for image 2. I think if added some saturation and contrast, the shot could look just as good as the first.
As for Gigapixel Al, I have tied it twice and didn’t believe it was worth the cost.
DxO has Canon R5 files now, so support for the Sony a7r4 hopefully will be coming soon.
Nice work.