Apply / Current Settings to All is a nice idea.
I can select for example enhance resolution, graphics, sharpen, standard and so on.
And then I can apply the current settings to all loaded images.
But my problem is the following: Every image is different in terms of how much sharpen strength / sharpen clarity / enhance suppression noise / enhance remove blur is needed for the selected feature. The auto detection is pretty good (when I load every image separately). When I use apply current settings to all, it also copies all of these slider values over to all opened pictures (even if I didn’t even manually touch the sliders and they still show green meaning detected default value). As a result it applies all values (from the first opened image) and applies these to all other opened images, which makes this batch feature unusable
Is there any way for me just to copy the selected filters / models (without their auto-detected slider values)? There are of course sliders (which are not auto-detected) that are important to copy over, like strength of face recovery (as they are just user preferences). I’m just talking about not copying the auto-detected ones.
If not, could you possibly add something like “Apply / Current Settings to All (without auto-values)” to a future version? So that these are still auto detected for each image?
That would really help.
Steps to reproduce issue:
Open multiple pictures
Set some main features like enhance resolution, graphics and/or sharpen standard
Click on the 3 dots on the image in the thumb view and select Apply / Current Settings to All
Now click on any other image (already loaded)
When you compare detail values (from the sliders) you will see, that all slider values mirror the values of the first loaded picture the settings were applied from. It does not auto-detect the optimal values for the other images.
As a result only the first image will have its best look. All the other images will look worse than they could (as the values were not really ment for these images, but for the first one)
P.S. I would have loved to use CLI as an alternative, but so far it lacks the ability for the user to set specific models (e.g. enhance=gfx, sharpen=std and so on) and will force autopilot all images.
Addendum: It is no fun having to do this to every picture manually in order to get the auto-values.
I just had to do this for about a hundred images. I had already collected them in one separate folder and loaded them all into Photo AI.
But then for every image: Clicking on the image in the thumbs view, then you have to wait for the UI. And while it is loading (detecting subject) the layout also changes so that the on/off buttons change positions. Even after they reach their final positions they are still not directly clickable until all loading has finished (detecting faces, …). Only then you can click on them or expand the models and select what you actually want.
It’s no problem for a single image. But for hundreds of pictures this makes you go mad
So I very much hope, that this is (or will be) possible via batch
Thanks! An idea (from a developer point of view, since I am one *g).
You could add a bool flag for each slider representing “the auto-generated value is used”.
As soon as someone touches the slider (changing the value or not) or types a value (even if it is the same value), the checkbox clears. If one wants to use the auto-value again, all one has to do is click on the checkbox again.
The big advantage would be that for “apply changes to all images” you would then know exactly which values the user actually touched and wanted applied to other images. All sliders still having the would not be copied along.
Currently you have the green dot there indicating auto or not, and double clicking on the slider to reset to auto; but an explicit checkbox would additionally hold the information “value was manually set and is therefore relevant for copying”.
This was already done in Gigapixel, the problem with Photo is the hover/thumbnail preview is the worst interface for this. GigaPixel has an alternate row/column spreadsheet style interface so that when you select a subset of images then select the subset of auto model or auto values or manual values that you want to change for that image set , it easily bolds/colors the rows/columns that change and a few quick scrolls you can verify before save that you have the right changes to the right images.
Windows has long used both styles of views, thumbnails and row/column details because both are better at different tasks.
I have already requested they port that UI option to Photo, as it already went thru the refinement process of user feedback so that it suited their batch workflows.
If you have not used that interface I highly recommend it, the old tools are cheap bundle adds to photo if you do not have it. That statement also applies to the devs, because the dev responded to it previously had clearly never used gigapixel in that way.
Hello @kevin.shank-0439, such a view would indeed be the icing on the cake.
But i’d settle for being able to copy only the main operations (without auto-values) in one way or another, since that would make my days much easier…
But thats the problem with doing it using the existing filmstrip and copy popup. While you want this model with auto params, someone else wants to set the sharpness slider, someone else wants to change the resolution. You basically have to reimplement the options dialog on the popup, yet even if that was done?
You are still left with the film strip being the worst interface for hovering, selecting, scrolling one by one, then hovering again to verify your subset options got applied to the subset of photos - compared to the detailed row/col view.
I spent over a week trying multiple times on several thousand photos just trying to turn off autoblur (what denoise actually does). My computer would lock up had to reboot/restart, or I would save and come back the next day and realize something was missed and still blurred.
I used gigapixel and while its interface was very easy to do this in mere minutes, turns out the AI models are not updated since January and lack the photo AI tweaks since.
I agree that that would be the best thing, but I am also a realist. Minor changes like tracking, which option was actually “touched” are pretty easy to implement (one checkbox in the UI and one property in the underlying code and a bit of logic). And modifying a method, to just copy over tracked properties is also done very quickly. That’s something you can easily achieve in one sprint.
Replacing the UI alltogether will take longer, that’s more on an epic scale. It will also affect UI Tests (if you have them). Plus you may have to convince product owners / management when you change the UI alltogether. No doubt - the UI in Gigapixel was superior, you will get no argument against that from me. But sometimes a new UI is also a marketing decision / strategy, to show we’ve got a brand new product, not just a variation of the old one - you have to buy that… Its seldom, that there are actual UI experts making the decisions (for what’s actually best). I can’t speak for Topazlabs of course - that’s just from general experience in software development projects over the last decades…
Gigapixel has same image preview and options dialog as photo the spreadsheetview is optional toggle. all the programming design is done they just need add new columns.
Photo is sold for restoring old photo collections and the existing filmstrip ui is simply not useable for that purpose setting up large batch jobs yet is also marketed as replacing gigapixel.
I agree, Photo AI, in its current state, is not usable for large batches in terms of UI.
That was something the UI for Gigapixel, etc. could handle with ease (that’s why I gave your idea a vote, you can return the favor with this thread ).
Reasons: It takes forever to load all images, and even if that worked, you have no way to change the settings per picture (either you take autopilot as is for all, or you can simply forget it - as clicking on every image in the thumbs and selecting something individually is just not usable, it would take forever and you would crash the app in the middle of your 1000 pictures - and then you would never ever try again…) And even if you actually did manage to set up all options for 1000 images, once you clicked on save 1000 images the UI would be pretty non-responsive in showing the progress of the saving images.
But in my opinion it’s still two different issues. My (minimal invasive) idea was to simply alter the “Apply Settings to All” method (which already works) just in that way, that it only copies the stuff, the user actually changed (by choice) in the template image. Instead of copying everything (auto-detected). For me that would already solve 80% of the problems. Even with the old UI. Since the default scenario is that I have a folder with images, where I know exactly what I want to apply (sharpen standard, face 20, enhance graphics). And I want to apply just these to everything I just loaded. I don’t want to tweak any of the pictures (I even could with the thumbs, even though they are slow).
But the way it is now I can completely forget that. As I can’t even set “sharpen: standard” for all loaded, without copying over specific settings for the first loaded picture.
I’m hoping that TopazLabs will do something about it - anything at all. Your option would be best with the Gigapixel UI. But if they don’t want to go for that (for whatever reason) let’s hope they at least implement something along my idea. Otherwise I don’t know how we are supposed to be using Photo AI for batches at all. And continuing to use the old products (Gigapixel, Sharpen, …) can’t be the solution, either. As Photo AI undoubtedly produces superior results…
Just changing the popup menu does not solve the problem of subset batch photo settings using a filmstrip to select things. Instead you would have to subset load using the windows browser and subset save batches which defeats the entire point of batching. And even that does not work with venn diagram batch overlaps these need 4x those need blur off that need sharpen. Just changing the popup like that means it still cannot be used. My stuff takes days to run and has subset venn settings to do. Are you going to rearrange your windows subdir structure with every patch release to keep up with option tweaks?
All I want to do is load an entire folder in Photo AI , select sharpen standard, face 20 and let it rip.
Without having to click on 1000 pictures manually.
If I try that now with “apply settings to all” it would work half way. But it would use all auto-detected settings from the first picture and apply that to the other 999. Which makes it unusable.
That’s the main use case for me. I know which pictures are blurred (I sorted them out already in the explorer). The auto-detect feature of when to sharpen a picture (low/medium/high autopilot preferences) is not really working for me. It worked pretty good in Sharpen AI.
And once that set of pictures is done, there will be no need for me to run them through another version of Photo AI.
Yes there is. They update every week. Maybe they will fix the poor results of streaks and blurs next week that you had no choice but to accept before. Are you going to reorganize with windows then and copy it to a new source dir structure moving things around based on subset setting batches only to find out it did not improve like they said? Using windows explorer for subset batching is not going to solve this, even if you have the time and diskspace to do that directory rearragning, you still cannot do venn diagram subsets that way (these are 4x but sharp, those are 2x but sharp, these are 4x but blur - are you going to rearrange that dir every release?). And given how each release has a regression failure, that is just not going to work. With gigapixel I had a dozen overlapping subsets and easily setup my batch job and let it run the rest of the week
Of course it would be nice to save that spreadsheet config for CLI batching and even XLS/CSV/JSON editing. But that is a seperate topic. Gigapixel is very easy GUI to use so is easy to redo next week since I will want to resample some previews to see if subsets need to rearrange anyways. Doing that with windows explorer instead is not feasible.
I could agree with you, if the photosets in Gigapixel would actually be persisted (on disk or database). But if you loose all the settings every time you close the program (and you will have to close it for an update), it’s still a lot of work to re-assign everything for every photoset.
If I was faced with that scenario I’d personally solve that with EXIF or IPTC tags.
I’d tag the photos with whatever info I needed for a future run (e.g. “process;sharpen:standard;face:no” or possibly even more generic along the lines of “blurry;no faces”) and then it wouldn’t matter, where the photo was stored (which folder). That would have the unique advantage of storing the info directly inside the image itself. Not in some external JSON/XML/… - because as soon as you reorganize your pictures, you also have to update your JSON/XML/… You could even edit the photos with 3rd party programs and save them, since most programs preserve EXIF tags.
I’ve already written a windows service, that recursively goes through a directory. I’d just add a method that would read out the tags of each photo and some logic to determine the appropriate additional arguments for the Photo AI CLI and trigger the CLI for each photo.
I could do the same thing without tags and just with file names (e.g. suffixes with certain meanings).
But: The CLI is not advanced enough yet, that it would accept parameters in that depth. Currently it will just auto-pilot every image. But hopefully it will be possible in the future, somewhere along the lines of Photo AI CLI - allow arguments to influence settings
[An absolute dream would be if loading such EXIF tagged images in Topaz Products would actually be interpreted on load and it would restore the settings - and vice versa being able to save the current settings in the original pictures EXIF data, If one chooses to. But I’m not kidding myself. We’ll be lucky if they improve one aspect of what we wrote in this and/or your other thread. As long as it’ll be possible to run a large batch of images with at least the same base settings and not the calculated values of the first image, I think it’s going to be a great improvement to now.]
P.S: By the way, I don’t use Windows Explorer. I use Total Commander, it’s much faster for file operations and using its shortcuts / multi renaming / selection / search tooling saves an extreme amount of time…
I agree this would be a great option for PAI to have. I would also like the ability to exclude a certain model from being used. Mainly the Lens Blur. Ever since it was introduced, PAI likes to use it a lot. The results are horrendous. Standard or Strong give much more satisfying results. So to add to the OP’s request, have the ability to exclude a model and use the AI to autodetect everything else.
I just had to do this again, this time for (only) 61 images. But it was such a pain in the a**.
I had the same kind of images (photography shots) and autopilot would detect some as enhance std and others with enhance low res. But I needed low res for all images (otherwise they look totally washed out).
So I had to load all images into the UI, click on every image once again, wait for the UI to complete loading, wait for the UI to change positions, click on low res and then after 20min I was finally able to click on save…
Please provide a better way for this… (CLI with selected models OR preferences with ability to disable some models OR Multi-Edit UI capabilities OR ApplyToAll without auto-detect values).
This is giving me headaches.
