Feature Request | Dust/Spot/Scratch AI Removal

Since this is an all in one photo process program it should include a one click dust spot removal and maybe a power line removal too. These features are available in Luminar NEO.

Photo AI does not seem to cope well, without help, with originals having “dust and scratch” problems, for example scanned prints.

This could be easily addressed by implementing something similar to the Photoshop / Elements “Dust and Scratch” filter.

The following example shows the problem and the difference that pre-processing through a “dust and scratch” filter makes. Here’s the original:

This is the version obtained by processing directly in Topaz. Although it makes a good job of overall sharpening and face recovery, the image still has a number of issues, most notably the dust on my father’s suit:

This is the result of passing the original through Photoshop Elements 12 and using the dust and scratch filter. No other adjustments were made in PS:

Finally, this is the result of applying default processing in Topaz Photo AI 2.2.2 to the latter image:

As you can see, it’s a dramatically better result.

Is there any existing setting which will make the equivalent fixes directly in Topaz? If not, could this be added as an enhancement?

Thanks, Andrew

I understand your use case and I have the very same one.
However, I doubt that TPAI aims at fixing issues like scratches though scanning.
You are better off fixing this via spot healing or clone-stamp yourself 1st and then hand it over to TPAI.

1 Like

The models have not been trained on scanned prints that contain dust and scratches and the application is not designed to remove those. The models have been trained on digital camera images.

You will need to use PS, AP or another pixel editor to remove those.

It feels like there are three options:

  1. Declare that Topaz Photo AI doesn’t handle this, i.e. it’s for “digital photos” not “digital images (which may include scanned prints)”. That’s the easy way out, but feels wrong.
  2. Add an optional pre-processing stage which can be switched on manually and just applies something like the Photoshop filter (which has been around for at least 15 years and must be pretty widely understood) in front of the others.
  3. Train the models on some scanned print samples. Some of the other “rescue your photo” AIs out there clearly have this capability so it must be possible. I can share some samples if required…

I’d be interested to hear from Topaz whether the product is expected to handle scanned images.
Andrew

I went ahead and moved this to the ideas section of the forum since this is a popular request.

1 Like

Dust spot removal is a feature of love to see implemented via AI. Thought I use the software on scanned images, dust spots are increasingly apparent at higher F-stops on my digital cameras.

Whatever software company can effectively implement this feature first, wins imo

1 Like

[Please describe your idea and how it would be useful to you in your workflow.]
Take a Photo
Add Feature For Remove Dust and Scratches.

1 Like

Thanks for the idea and suggestion! Make sure that you do not forget to vote for your idea :slight_smile:

Are you referring to dust or scratches that may be present on the lens or sensor or do you happen to work with old scanned images?

Both John

2 Likes

Add Focus Stacking.

2 Likes

All these responses seem to miss the point, there are other image AI editors (Luminar Neo for example), that remove dust spots automatically. So yes, it is possible and no it doesn’t require someone to spot heal all the dust. I’m torn on upgrading to Topaz Photo AI or going with Luminar Neo. I prefer Topaz as a company and their software is IMO better… but I really need the dust feature. I can’t afford both. Automatic dust removal is a god send when you’re processing 100+ images and have dust on the sensor. Manual spot healing just won’t do.

Many scanners have IR and normal light. The IR allows for dust to be isolated from the rest of the image as a mask. It would be amazing if you used this to train a dust and scratches model that actually worked to fix film scans. That alone would get you the entire film community. All older dust removal algorithms are garbage and I have tried most of them.

Also, bonus points for accepting tiffs that have the dust mask and actually using that as well. Probably not necessary but would be amazing.

I’ve tried Hip Paws Scratch Rermoval. Love It . Any plans to add this feature?

Thanks for the idea and suggestion! Make sure that you do not forget to vote for your idea :slight_smile:

It would be nice if you could give us customers a time schedule for when to include removal of scratches and dust from old and scanned photos in Topaz Photo AI.

1 Like

We currently have this scheduled for sometime in the future but it’s not on our immediate roadmap or priorities. If more votes or other indication that this is higher priority than some of the other stuff we’re working on, we may consider bringing up the timeline.

Hi Tim,

Thank you for the reply.

You have very high quality on your products, so I think you will put heavy efforts into the subject when prioritized.

Its worth waiting for :slight_smile:

Best regards,

Olav W

tor. 14. sep. 2023 kl. 20:48 skrev Tim He via Topaz Community <noreply@topazlabs.com>:

1 Like

Managing dust spots on scanned prints, as well as sensor spots, is a big deal in the genealogy field. Everyone on Wikitree, Ancestry, FamilySearch, etc, would benefit greatly if LR trained on such images to remove crud without over-polishing the old photo.

2 Likes

another use case - I make photos of paintings, some are protected with scratched plexigalss which is often glued to the painting.