I have noticed and DeNoise AI and Sharpen AI significantly increase a JPG file’s size.
For example, if i take a Nikon raw file (NEW) from a D850, the native size is about 53mb. If I process it in ACR and export it as a JPG with no size reduction and at the highest quality, it will be 30mb or so. Running it through either DeNoise AI to Sharpen AI will increase the size to 53-57mb when saved back to the JPG. I can then take the extra step of running the image through Photoshop or OnONe Resize, and the image reduces to 33-35mb with no visible loss of image quality.
I have had files processed through Luminar come out as big as 95MB after processing through Sharpen AI. Luminar does not seem to do a good job of compressing JPG, either.
Can this be addressed? It can be a problem when going over the limits of some file sharing sites.
Part of the Save As dialog is a quality slider that let’s you tune the amount of jpg compression being applied. This provides you with some amount of control over the resulting file size. (at the cost of image fidelity)
That doesn’t explain why 100% quality in PS or OnOne results in a much smaller file than 100% quality in Topaz.
Hi Andy and Mick
I use DeNoise AI version 2.0.0. and I only find a “Save” button. I will probably look in the wrong place? Hope some one can help because I would like to keep the image file size under control.
All the best
The best method I have found, which also preserves the EXIF/IPTC info, is to open the file in Photoshop then open it in DeNoise AI using the PS Filters menu.
Once you are done in DeDoise AI, the image is returned to PS which will save it in the usual manner and at a reasonable size. And, as I said, the EXIF data is not messed up.
This also applies to AI Sharpen.
Let’s for arguments sake say a TIFF file with no compression contains 100% of the available pixel detail of an image. That then begs the question as to whether any JPG copy of that image, which has had lossy compression applied can be truly 100% quality in comparison to the TIFF.
I take it to mean only that 100% quality is the best quality that particular export process can provide for that format, and that any other percentage setting below that in that particular export process is in comparison to its 100% setting.
I don’t take it to mean that one program’s 100% is the same as another programs 100%.
This is a major issue for me - my work sees me typically submitting dozens or hundreds of images to my clients. I export 20Mpix images from Capture One at around 5-8MB per image. After running them through DeNoise AI they come out - looking fantastic, it’s true - at around 21MB per image.
This is with setting the output quality to 7 …
Surely some mistake …
If you can run DeNoiseAI or SharpenAI as a plug-in from a host application, then save the file from that host application, the file size r and the EXIF will be as the host application would do normally.
You just run into the problem when using the Topaz apps as standalones.
Thanks mickklass - I’ll give it a go. Still seems like pretty rogue behaviour though …
Interestingly, it seems if I export images from C1 as TIFFs and then run DeNoise on those then the resulting filesizes are much more sensible!
(C1 doesn’t seem to play well with DN as a plugin).
Here’s what is happening –
If I choose an original file of 4.6mb and run it through AI and then save/export the file – one of two things are happening.
If I save that 4.6mb file and the Format is set for “Preserve Input Format” it saves that file 4.6mb after processing at 2.0mb. Less than 50%
If I save that same 4.6mb file and Format is set for “JPEG” it saves that same 4.6mb after processing at 5.2mb. Actually, increasing the file size.
Obviously, each file is handled differently, and end results will vary per each process, but the difference in choosing your export Format as “Preserve Input Format” versus “JPEG” is significant. Users are not realizing the difference and are complaining on message boards, etc. that their file size is decreasing. I am not sure what the Format “Preserve Input Format” does differently, but if the user is assuming they are running a JPEG through AI and they choose “Preserve Input Format”, that they do not have to specifically chance the output Format to JPEG to maintain overall file size, they will have this problem.
Going from 90% to 100% on the output slide makes the change to maximize file size, but the real problem is the perceived “Preserve Input Format” makes a significant decrease in file size on export and people don’t realize that the only way to preserve file size is to go into the drop-down menu and specifically request JPEG for export.