I don’t regularly change my Prefs after setting them initially… I just checked.
I believe this is the one I was wondering about & whether it might impact how you’re seeing your image colors… I think you can also set a colorspace when saving. But that’s later in the processing than what I understood your issue to be.
As you can see, I keep that Pref turned off. Perhaps you can experiment and see if it makes a difference for you. If no help, you can always flip the switch to “off” again. In which case, there may be some other issue the Topaz team is aware of and is addressing. I’m just a user. Good luck!
There are so many posts and activity here it’s impossible to keep up with, but it’s nice to see a thriving community here that cares so much about the product. That’s a rare thing!
The biggest opportunity I see for PhotoAI to help me out are with old scans of photos. I have thousands of them, and I simply will never get to editing them one by one in Lightroom (I know because it’s been 10 years of me thinking I might someday do that).
So what would be tremendously helpful would be features that would:
Clean-up of dust/scratches
Red-eye removal
Auto-crop
Auto-rotate of landscape images in portrait orientation
Sounds like interesting ideas if you’re using PAI as a full processor vs as a pre-processor.
I’m just curious (as another user…), how would you expect a program to auto Crop for you?
How would it know where made sense (that you’d be happy with) for different images?
What parameters would you want to give it to batch auto Crop (for ex., leave x amt of space around subjects, leave more space at bottom than top or sides of subjects, etc.). Would you only expect it to be used for studio portraits vs environmental portraits, landscapes, architecture, travel photos, still lifes, etc.?
Yes, I have 3,646 scanned images and I’ve accepted that I simply won’t make the time to process them all one at a time. Automation would be a big help here.
On the cropping, I’d only want it to crop any thing outside the photo that the scanner picked up. Should be relatively easy to pick up when the software is looking at the image. And auto-cropping is the least useful thing on that list to me.
All the scanning happened via a small, portable scanner designed for bulk scanning. And the scans were mostly done by someone else. No opportunity for software to do anything to help. But also, scanning 3600 images via a scanner on a desktop would be quite the tedious task.
It’s not helpful in the least to tell me what I “should” have done.
I have kind of the same situation with my own film and print scans.
One possible solution is to only edit them when you need to! ie, if you’re going to post some on a blog, send to friends or relatives, etc. Otherwise all of those hard-work-edited photos will probably just sit on the hard drive unseen anyway.
This is not my specific answer to the original question, but as for removal of unwanted image surroundings… Sometimes I found it easy and quick to remove the background. But I only have maybe 50 such relevant images with backgrounds (and not from a scanner, they are photographed photos). Taking thousands of photos manually would be a cruel punishment .
For batch processing (including also other functions) I know of the following tool (but it is online). However, contrast also matters, so it’s not 100% reliable. I guess there are other similar tools, too, maybe better (but probably costly):
Yes, sometimes (like in this case) it has to be edited/finalized manually. I don’t have a better example at hand right now. If the image is scanned, then the properly inserted original does not suffer from disturbed perpendicularity of the edges. The authors of the photos should think much more about us, who then edit it for our own pleasure !
Yeah. I doubt those other programs would be able to do face recovery and things of that ilk. They just specifically address the other features you mentioned in your wishlist.
That’s why I suggested that a program of the sort I mentioned (which gets excellent reviews from what I could see online…) might be a ‘stopgap’ for you until/if the Topaz folks were interested in adding those sorts of additional features.
In the short-term, it would likely require you to run 2 batch runs. One to accomplish what PAI currently provides for you and one to address your supplemental features (rotate, crop, clean scratches, etc.).
A deja vù again - this allegedly faster version V3.5.0 is in fact slower here on the M2 Ultra than the previous V3.4.4
So here are the times (Super focussing a standard iPhone image at 4032x3024 pixels): Sharpen strength always set to “low” (but didn’t make any difference here)
Focus boost none:
V3.4.4: 4m 02s
V3.5.0: 4m 58s
Focus boost minor:
V3.4.4: 2m 05s
V3.5.0: 2m 30s
Focus boost major:
V3.4.4: 39s
V3.5.0: 46s
I noticed that the new version has a higher power consumption on the Neural engine but lower on the GPU - which again, as seen before in Gigapixel and TVAI hampers performance on machines with high GPU core count, at least for the M2 chips. (It does also reduce the total power consumption with the V3.4.4 taking about 140-150W and the 3.5.0 110-115W).
Obviously the Mac tests are done only on lesser equipped chips - otherwise those claims of a speed improvement couldn’t be justified.
So: please, please make a setting in the prefs if the rendering should be done preferably on the GPU or preferably on the ANE.
Otherwise users of those high-specced ultra CPUs are forced to stay on the older versions with both, Gigapixel and PhotoAI (and thus kinda have to quit their subscriptions)…
Why don’t you pick out the best / favourite 100-200 pictures and do say 20 -25 per week and then after that the next best / favourite 100-200 pictures. You should find that you develop a routine and it becomes easier / faster. I know I’ve done it. Otherwise this time next year you will be looking at the same pile of pictures and thinking “I must get them done sometime”
Do you have similar choices? Have you tried all 3?
Ideally the Auto choice should choose the best way, but for years Topaz seems to have a broken Auto setting. I also reported problems with it for Denoise AI, Sharpen AI, and Gigapixel AI years ago on my Windows PC. I found at that time that often Auto was slower than specifically choosing the GPU and sometimes choosing the CPU was faster. It depended on which model, which app, etc. Hard to pin down. But sometimes Auto would be the fastest.
Recently someone posted here and said that his Apple Silicon CPU efficiency cores were being used, but the performance cores were not. I checked and discovered the same thing when set to CPU. Strangely, I also found that when set to CPU that the GPU was also being used a lot.
It seems to me that the Auto setting needs to be improved so that it uses the best combination of CPU, GPU, and Neural Engine for each Apple Silicon chip. And, I suspect, similar problems with Windows PCs. It seems not to be smart. It would require benchmarking for all the M series chips and by now there are a lot: M1, M2, M3, M4 series with multiple variations (standard and binned) for each. Maybe this is a place where AI would help?
Years ago in my emails to Topaz support about the Auto setting for AI Processor I suggested that they could add an option for the user to run a Topaz benchmark that tried various combinations of CPU, GPU, and Neural Engine. I sent Topaz a link to the Neat Image noise reduction program that has a very nice implementation of that to fine tune it for each user’s computer.
In Neat Image go to Tools | Preferences | Performance and click on Optimize Settings. It will run a denoise benchmark and automatically determine whether using the CPU, GPU, or a combination will give the best performance on your computer. And it also determines how many cores to use for best performance. You can then accept its results or choose something different.
If Topaz is unable to get Auto to provide the best performance then maybe something like this could be added.
On my Mac Studio M1 Max, 64 Gb I tried to use “Superfocus” …… after normal pictures I tried to use it with a picture of 68kB… even with this small picture it did not work at all….the process hung, I had to force quit TPAi
In which case, there may be some other issue the Topaz team is aware of and is addressing. I’m just a user. Good luck!
.
!
