This week’s update addresses some of the feedback we’ve gotten since release (and more is on the way!), fixes numerous bugs, and introduces a trial mode so you can try before you buy!
If you own DeNoise, Sharpen, and Gigapixel with up-to-date licenses then you already have access to Topaz Photo AI!
Please give us any feedback or report issues with this release. We’ll be updating TPAI regularly to address those pieces of feedback and issue reports. If there’s a specific image you’d like us to see, you can send it us at this dropbox link.
Updated various camera profiles to have proper exposure baselines, including logic for determining camera make/model, and adding several new cameras
The windows installer now checks for AVX2 support before installing
Removed padding around the app icon on Windows
Filters are now hidden when refining subject or selecting faces, and the save button is disabled
Fixed “COne” folder being created on computers with multiple hard drives and don’t have Capture One installed
Fixed some images having incorrect cropping
Fixed Mac Capture One plugin not working on files with spaces in them
Fixed auto-upscaling showing check when using plugin mode that doesn’t support resizing
Partial fix for crash on some monitors with custom colorspaces
Partial fix for window being larger than the current monitor’s display
Going forward we’d like to share a bit about what major feedback the team is actively working on addressing, to give an idea on what you can expect in the coming weeks. There are no promises these will be completed and ready by any specific release, only that they’re in the pipeline. Here’s the current list:
Subject refinement is not precise enough
Autopilot takes too long looking for subjects or faces, particularly in cases where the user knows there aren’t any
@anthony.lawn Just real quick… Did Ya’ll change the memory model or something??? On my M1 Mac w/16G of ram, processing a NEF (RAW) image via the LR plugin… if I select the 4X upscale it processes okay (still slowly) but saving back as a DNG file crashes/fails to complete because it exhausts Application RAM…
Not in this update, no. When you change versions M1 devices do need to rebuild some caches though so the first time you use each model will take longer than normal. That may be the cause of the problem, although I wouldn’t expect that to cause it to run out of memory when it was fine before this release. If you’d like to contact Topaz Support, they’ll help get more information about your machine that could help us fix any issues that could be causing this behavior.
I have an additional input. You have to wait for auto pilot to process then change the resolution and wait again to process. I’d like to be able to set the scale before processing.
A.P. removed the chromatic aberration okay along window edges in city bldgs using the raw denoise (using my unprocessed Sony ARW raw image).
A.P. detected single tall tower close to center of image as the subject on 'Default’.
I switched to ‘Landscape’ using Refine - Refine expanded to grab all the buildings and the park in the foreground. I really only wanted to select the buildings to Sharpen more…
I couldn’t find a way to do that. I tried various slider pulls with the sensitivity and hardness/softness. Couldn’t adjust to include only the bldgs. - Also, I didn’t see any processing indicator for the Refine so wasn’t sure if it was thinking/processing or had thought and couldn’t do what I wanted.
Split View - Removes Chromatic Aberrations & Color Noise along window edges (a gd thing):
1.0.1 plugin launches from inside Ps 2022 host. BUT, is back to launching centered over Ps UI vs filling the screen as it should… Pls. can that be corrected.
Can we pls get a File > Prefs ability to set to either Sharpen Subject Only or Not (an On/Off type setting)? Then we can use the switch (under Sharpen) in the righthand Settings Panel to do the opposite of our Pref if we want to override it. I find more often than not I do not want the Subject Only sharpen option b/c the masking is so funky… until that masking is more controllable I’d want a sticky Preference setting to turn off Subj Only and only turn it on selectively if able to create a desired mask…
Also for the File Menu - can we pls get an Open Recent file option? (I realize there are more fundamental fish to fry right now… but that would be a good to have on the roadmap list - it’s pretty inconvenient to have to keep finding images again).
Thx. p.s. That’s wonderful to know what’s coming in the pipeline wrt fixes!
I really like the image quality photo AI offers so far. The results look much more natural than the separate programs.
Now that the monitor color space bug has been fixed, it seems there is one more bug in regards to color management. I noticed that the saved images from Photo AI don’t display properly in the Windows Photo viewer (the legacy one, which many still use in lieu of Windows Photo because it is color space managed properly). The image looks brighter and flatter than it should, as if it a linear tone curve encoded image.
There are a couple of little things that would be really nice to have as well:
When saving a TIFF with LZW compression, Photo AI currently only offers 8-bit output. It would be nice to have 16-bit output with LZW compression. Sharpen/DeNoise/Gigapixel AI can do this, so not sure why this was dropped from Photo AI.
Offer a choice of output color space for JPEG and TIFF output. Sharpen/DeNoise/Gigapixel AI can do this. Furthermore, it would be nice to have more choices for output color space. In particular, Display P3 (DCI-P3 but with D65 white point and sRGB tone curve) would be a nice additional option that the older AI programs do not offer. Even better, allowing custom selection of output file color space would be great. This could easily be implemented using LCMS as part of the program’s engine.
Masking features, especially if there can be separate masks for sharpening and denoising. The cropping in gigapixel ai would be a welcome feature too (I’m adding to my post at work , forgive me if that’s already there in photo ai. )
This would all be covered when they provide an option to disable autopilot on image load! I will keep asking for this and hope they will ‘see the light’ as to why we need it.
Then add batch processing, masking tools and we will be close to perfection!
Subject detection is still not good as detecting a motorcycle as a subject rather than the person sitting on it is a strange choice. The only way to select the person (as well as the motorcycle and deselect the tree) is to Refine with Portrait instead of Default selected … seems a little illogical:
Loss of detail also happens often. See the RHS where the yellow in the flower is lost after processing…
I downloaded v. 1.0.1 tonight. It goes through the processing, updates the preview, but then freezes. I can’t save the image (preview looks really good though!). This is on an iMacPro (2018) running OS Monterey v. 12.6.
Topaz should add an menu option to turn OFF the automatic mode. They can leave it ON by default, and let us decide if we want to turn it OFF. I think that’s the best of both worlds.
.
