A common thing that needs to be repaired in landscape photos is highlight blowout ( true blowout area as in all white with zero underlying image data to recover) in the clouds. I can’t find a way to do this and there are no prompts that can accomplish this that I’m aware of. I would like to keep my clouds and NOT do an entire sky swap, but simply generate and swap out the blown out area within the visable non blownout parts of the clouds, filling the blowout with a regenerated, blended similar cloud type thus keeping my photo as original as possible but eliminating highlight blowout.
At the moment, there isn’t a way in Topaz Photo to selectively regenerate true highlight blowout areas (where there’s no underlying data) while blending them seamlessly into existing cloud structure without doing a broader sky replacement. You’re also correct that there isn’t a prompting mechanism available currently to guide the AI in that specific way.
That said, your idea of localized cloud regeneration, filling only the blown-out regions while preserving the original cloud formations, is a great one, especially for photographers who want to maintain authenticity and avoid full sky swaps. We’ll be sure to share this suggestion with the development team as a feature idea for future consideration.
Really appreciate you outlining both the problem and the ideal solution so clearly. Feedback like this is exactly what helps guide longer-term improvements.
1 Like
So many of my landscape/sky photos from early digital cameras from the 2000 to 2010 had blowout prone sensors and to this day even with cameras over the past decade I still have to severely underexpose or exposure stack with multi shots to try to rid my photos of this persistent, patchy dataless evil. I have hundreds if not thousands of photos thus affected that I never deleted in the hope that one day they can be salvaged. So this type of repair, combined with your current upscaling, could help me bring to market many such old, but well-composed, comparatively tiny-1600 x 1200 sized blown-out photo files from first digital camera right up through my newest modern sensor 6000 X 4000 shots
1 Like
This is actually great idea and I believe should be very doable. I shot a lot of of landscapes, usually trying to either have multiple exposures or underexposing a bit to compensate. However it still happens too often there is (or even few ) area(s) of sky with bits and pieces of clouds overexposed beyond recovery. Sometimes if photo is worth it and I cant accept these overblown areas I am “fixing” them in photoshop with variouds methods based on the surronding clouds. Having this process automated would such a time saver at times.
Yes, I hope that Topaz can do this. I find that Photoshop’s generative fill does not do this well and some Photoshop repair methods just grab a piece from elsewhere in the photo, risking an obvious repeated cloud section, which I would hate even more than a blowout. Thus, a procedurally generated or AI fill-in-the-selection method would be preferable. Another method of repair via Photoshop, staying fairly true to the original scene, is time-consuming but involves grabbing a non-blown-out cloud piece just out of frame on a cropped image or from a separate wider-angle shot of the same scene at about the same time to avoid repeat parts. But that takes a lot of time blending as well.
1 Like
Highlight /shadow warning
Less sensitive sliders look DPP Professional 4 Canon
Wonder is very slow even on very high video card and proccessor
Thanks for sharing these ideas — we’d love to understand a bit more so we can pass along clear feedback to the team.
For the highlight/shadow warning, are you thinking of something like clipping indicators (blown highlights / crushed shadows) similar to what’s shown in other editors?
For the less sensitive sliders, could you describe which adjustments feel too sensitive? If you have an example or comparison to Canon DPP4 behavior, that would be really helpful.
And regarding Wonder performance, could you share a bit more about your setup (GPU, CPU, RAM) and what kind of images you’re processing? Also, roughly how long is it taking per image?
Hello John,
Thank you for responding so quikly,
Clipping indicators that would be great
Edited raw photos from Canon DPP4 lose their editing when opened in Topaz, which is a shame.
This could work together better
For the sliders a slide that follows the steps of underexposure end overexposure, as this occurs in a camera.
I tried the wonder stand online with a jpeg photo, wich was mutch faster, but the result is disappointing in my opinion.
I have Amd Ryzen 7 5700 3D cpu with 32GB DDR memory and AMD 5700XT videocard 8GB.
I hope you can still improve some things, that would be great.
Regards,
Raymond
Verzonden vanaf mijn Galaxy
-------- Oorspronkelijk bericht --------
