Every image I process with the latest Denoise AI 2.0 produces a weird color shift and total loss of contrast. The image shown here is typical, and looks like this as soon as I open it in the application. What’s wrong?
2 Likes
I and several others have also noticed and reported this effect. As it was explained to me, this is occuring apparently because I’m using RAW images and the libraries Topaz uses for Raw processing haven’t yet been updated yet to work with DeNoise. I was advised this is on one of their priority lists, but I have no way of knowing just how high up on the list it is.
There is a work-around which I’m using in the meantime. After doing the DeNoise processing I pull it into Photoshop. I then go into my adjustments windows and select the Curves adjustment. If you haven’t used this before, you want to make an “S” curve instead of the 45 Degree straight line. Take the cursor and click on the 45 degree straight line at some point below the horizontal center line in the window, then drag it slightly to the right and release it. Now click on the the diagonal line above the horizontal center line and pull it to the left. You will need to play with each of these a little to get just what you want.
1 Like
Joel didn’t say he was using RAW files and his color shift was more than a curve correction. Normally, you would expect that going from a photo program (photoshop, etc) to Denoise AI would not need any correction and is the best work around. I don’t know if that is his case or if he was going directly into Denoise with a RAW file. Maybe he can clear this up.
Yes, it is a RAW file opened directly in Denoise with no corrections at all.
I first tested by exporting original (RAW) files out of lightroom and opening in Denoise AI 2.0. Found I have the same issue. The color profile appears to be stripped out and the change is visible upon opening the file in Denoise AI. These were RAW files with sidecar XMP files.
I tested a bunch of variables for the export format (RAW, DNG, TIFF, JPG) from lightroom and found I could retain proper color if I opened a TIFF or JPG in Denoise AI 2.0. The “save as” format didn’t seem to matter. So until they get this straightened out I am opening TIFFS in Denoise AI 2.0.
I’m getting this too.
Hi Artisan-West, I realize he didn’t mention he was using RAW images, but since it was exactly the same symptoms I was getting when using DeNoise I figured it was a pretty good chance he was. I’ve posted a step by step procedure I use for my temporary fix. It was Adam, one of the developers I think, who gave me this advice. So I didn’t come up with it on my own. I’m not that brilliant.
Here is a step by step procedure I’m temporarily using until Topaz updates the color libraries.
Here is my original ARW (RAW) image. Be sure to click on the images first to see the entire image.
Here is how it turns out after processing through DeNoise.
Here is the Adjustment window which you need to open up. If you don’t see it go to the Photoshop Window drop down column at the top left of your Photoshop window.
Now click on the curves adjustment.
Now play with the curve to make the image the way you want it to be.
I would normally go into Adobe Camera Raw first to make exposure, shadow, and light adjustments, which would make this image even better, but didn’t do that in this example.
1 Like
Lightroom is a non destructive processor. This means that the adjustments are not applied to the picture except in Lightroom or when exporting. I think that if you jump from Lightroom directly into one of Topaz’s filters you are getting the RAW unadjusted picture (I could be wrong since I haven’t used Lightroom for a few years). Using Photoshop or Affinity Photo (with a stamped layer Ctrl-Alt-shift-E) adjusted picture, go to Denoise AI, etc. directly from a filter menu which should produce the best result.
I have it, but I’ve never used Lightroom before. I might try your suggestions as an experiment. According to Adam.mains (I was wrong he is not a developer but one of their software engineers, they do know about the problem and the solution but he said it might take a couple more BETA upgrades before it is fully resolved. I hope so, I’ve got just over 5000 photos I want to send through it and batch processing, for me anyway, is well worth waiting for. Sending that many photos through DeNoise one at a time is a pain. I’ve done it many times before, but now that I may not have to do it again is a really good thought. I do understand why most people would find very little use for it though.
I foolishly updated today before properly checking for posts like this one.
Has Topaz given any kind of timeline for .raw support? Or a way to rollback if it’s a ways off?
1 Like
If you still have your .exe for the earlier version that would be the only rollback I’m aware of. Haven’t heard of a timeline for the .raw fixes, but I don’t expect it to be a very long wait. A few weeks would be my guess, but that’s all it is.
I did the same thing. Hope they respond to this soon.
Any updates on this?
I purchased the product a month ago and still have not been able to use it as intended.
Miguel, I have seen some good improvements. Check to see if you have the latest update which is version 2.0.0. This version is giving me what I need. Here are 2 examples. I took 2 RAW images and ran them through DeNoise AI and then saved each as a JPG image. I would normally save them as a TIF before beginning any other enhancements to the images. Obviously I will be processing them further but I wanted you to see them just as they are after processing with DeNoise AI 2.0.0. Click on the second photo to see the entire image.
Miguel, I decided to finish the processing on the second image so you can see a final product. Again click on the photo to see the entire image.
Thank you Terry for your detailed responses.
I am new to DeNoise and love what it does with noise, I recently started taking pictures of indoor sports at my children’s games and noise is a given.
I shoot Nikon RAW. I am new to DeNoise and still trying to figure out the best approach to a workflow. I read that DeNoise should be applied early, perhaps do some minor exposure adjustments in Lightroom which is what I use, export to DNG (or RAW), run DeNoise and then bring back to either Lightroom or Photoshop for finishing.
If the above approach is correct then I would like to bring them back from DeNoise in a lossless format, either DNG or TIFF, but it seems that the only way that I can bring them back without the “washed-out” effect is to save them as JPG.
Does the above approach make any sense? I would greatly appreciate any feedback. I’ll post the results of this morning’s test in another reply.
Terry,
This is what I did this morning.
Took a NEF image in Lightroom, adjusted exposure a bit, exported to DNG
Brought it into DeNoise, used Clear-Auto and exported as DNG.
This export was affected, it looked flat and colorless. I brought it into Photoshop and worked on it to a level that was satisfactory to me. My gut feeling and perhaps I am wrong is that we should not have to do much to the image to bring it back from it being affected by DeNoise, but I sincerely welcome a different point of view.
The images are in this order, original DNG export from Lightroom, DNG export from DeNoise and DNG after being adjusted in Photoshop, I saved them as JPG to save loading time, but I think they closely represent what the DNGs looked like.
PS: I have to figure out how to embed images in a message, I tried uploading the files and received a message that I could not insert files in a message. My apologies for not having the images embedded.
Miguel, I attempted to access the photos you sent, but I received a notice “URL not found.”
To send me (or any other person on this forum for that matter) a photo, just hit the “reply” button and when the message window opens up type your message. When you want to place a photo into it, just set your cursor to where you want the photo to go and drag and drop the image file into the message box. If you have Photoshop/Lightroom you may want to change it’s size to 72 dpi and maybe 12 - 15 inches width. If it is too big, this message box won’t take it. It will quickly become part of your message and it will go along with your text to whomever you send it to.
Did you check to make certain you have the latest version of DeNoise? You can check that by going to the “Help” drop down at the top left of your DeNoise window. Go down to “About” and release your button. It will tell you the version you have installed. It should be 2.0.0.
I try to do nothing else but DeNoise AI to an image outside of some exposure adjustments, before I do any other adjustments. If I don’t, I most likely will exacerbate any flaws it might have.
The above approach you mentioned is what I do as well. However, when I have a great many, as I’m about to do now with 5000 images, I will be batch processing them with DeNoise, and save as TIF’s. No difference in quality than DNG, I’m just used to TIF’s. You may find differently but I find than I can process all of them at the same DeNoise AI setting whether they need it or not. It saves me a ton of time. Examining all 5000 of the images I’m about to start working on to see which ones need and which ones don’t need noise reduction would take many hours and for me would be a big waste of time. I’ve come up with a DeNoise setting that I like and that does an acceptable job with most of my images and that’s the one I use. Since I’m saving them back to the same folder I can go through them all after they are processed and if I happen notice one that didn’t turn out right I can do that one over with custom settings.
I access the DeNoise filter from within Photoshop after creating a new layer (Ctrl J). When I save it, of course it comes back into Photoshop. Then I pull it into ACR and do my basic beginning processing adjustments there. I have changed my default settings for ACR so that when I pull a RAW photo into it, ACR will make the same basic settings I use for nearly all photos automatically when it gets taken into ACR. I can scan them and if I see an image that needs adjustments I can individually adjust that one.
