DXO PL Deep prime and Vuescan

Deepprime is designed to keep small details uneffected wile sensor noise is dealt with.
So Deepprime has a sharpening effect due the fact that the thin lines are clearer visible.
So small structure like brickwork on a distance is less effected.

I used to use negative scanners and Vuescan but now I just use extension tubes and photograph the negatives over a light box. I use negative lab pro to process.

Using electronic shutter and app control allows taking multiple photos which can be merged to reduce sensor noise very effectively.

The results of this (with a Sony aps mirrorless) are far better than with the Nikon / Minolta dedicated negative scanners.

True, there is no digital ICE but the scratches are easily removed with the modern healing tools which did not exist in the days of negative scanners.

Günter,
if reducing scanner (color-)noise is your main concern I recommend scanning your negatives as 16bit-TIFF and using the free and opensource “RawTherapee” for denoising the images without killing the film grain. You may define denoising profiles for different types of film as presets. Works fine for me.
Greetings
Kay

1 Like

Dear Kay,

thanks for the info. At the moment I’m waiting for the scanning results of the 3 slides I send to the company in Berlin. Then I will make some tests with software I still own. I will also try my old licensed Neat Image version :grinning:.
See what happens

1 Like

@Guenterm
Sorry to say but I wouldn´t count on Deep Prime or Deep Prime XD at all to solve your noice problems in slides. I have used Sony RAW reproducing a lot of color slices made with Agfa CT 18 and 21 and despite they certainly are in ARW the Deep Prime or Deep Prime XD won´t do anything at all with them concerning both the noice and either Unsharp Marsk or the new Lens Correction sharping.

The best tool to get rid of the sometimes-terrible grain especially in areas with uniform color like skies is to push the Microcontrast as far as possible to the negative side and instead use Microcontrast up to 35-40. That will increase sharpness and you can increase sharpness during export by using “Bicubic Sharpen”. Be careful with the later though because that method doesn´t look especially good if there is three leaves on the image because they tend to get overshapened. In that case select just “Bicubic”.

Your best bet I think will be Topaz because Topaz works even om JPEG I think, but not even Topaz will be able to handle the big ugly grain especially Agfa CT 21 produces sometimes. I tried a lot of tricks for a couple of years before I landed in the process I described above

I will try to use Vuescan för my color negatives if it works properly but for slides I will continue to use my Sony A7 IV and Sony 90mm macro.

I just was curious and tried to open 12 year old DNGs from Vuescan in Affinity Photo, worked. At the time I used DNGs I was just afraid the scanner mechanic would not last long: VueScan made the scanner rail move to each single negative it detected (and very often not detected). So I saved the whole flatbed area with the film holders as DNG and let VueScan find the frames in the RAW. 2010 only VueScan could open it’s own DNGs, especially if the additional IR scan was involved to detect dust on colour negatives. But these days even Lyn can open these files.

As for the noise reduction, film grain already is noise and some information just got lost by the size of the grain. If that noise could be filtered, the information still can’t become restored. So, I think it’s a bit pointless to use DP as it was made for digital noise. Film grain has a different structure, no?

At the moment I’m using a Fuji X-T2 with a Laowa 65 mm 2× Macro and a film holder from VALOI (Finnish company) to scan old 135 negatives, b/w and colour. It’s quicker than the V750 and generates at least as good, if not better results as I was used to get from the scanner. And saves the trouble to detect single frames (for VueScan).

Dear Sten-Åke, dear Joachim

thank you for finding this old post and paying attention to it.

I have now implemented many of your suggestions and I am proceeding as follows
Scanning the slides and negatives with a Plustek Opticfilm 135i in TIFF format
basic processing with DXO
as last step processing with Topaz products

And for the last step, I’m still a bit on trial and error. With the 3 individual Topaz products, I can use more moderate settings, run a batch of 400 photos, and rather sharpen and/or denoise a few more frames. With the new Topaz Photo AI, the batch runs through with automatisms and I have more to rework so far. Also, Topaz Photo has a problem with memory leaks, but which has been analyzed in communication with the very good Topaz support, and sending some log files. Probably there will be a patch for it this week.

In any case, I will process the slides of my friend and my slides and negatives this winter and then also put a line under the chapter.

I wish you a good time and a lot of fun working through your surely more extensive picture stock.

Until sometime

Guenter

(Prime?, ) Deep Prime and DeepPrime XD are meant to be used on bayer data, produced by digital cameras.

So data that still needs to be demosaiced (or the Fuji x-trans thing).
The DNG written by Vuescan is just like the output DNG created by DxO: It’s a linear DNG, that already contains R, G and B data per pixel.

DeepPrime and DeepPrimeXD will (probably) never work on this, they are not designed to be used in this kind of data.

@JoJu
Thank you for that info. I will also try to take even the negative color images with my camera and handle the color conversion in Vuescan.

Jepp, so it is.

@Guenterm
Well I feel it’s I who have to thank you. Infact it has been the color conversion of the negatives who have stopped me so far to process these color negatives and that I still have so many color slides still unprocessed but I will get there after some testing.

Intresting to hear about the steps you have taken to solve these problems so far Guenter. I hope Topaz will manage to solve the application problems you have. It can be very frustrating to have to rely on software that have issues.

So, DP could be useful if a camera plus lens is used to reproduce old film images? Still, even if so: I like the film grain. Although it’s more like flakes as I often used HP5 and developed it to 36°/3200 ISO. Yes, it is a bad idea, but either available light or flash and I found flashing more distracting than grainy pictures. Which left something/a lot/everything (depending) to imagination. :face_with_raised_eyebrow: :smile:

I took the Fuji to not buy a new lens for FF, the Laowa is at least better than both Nikon Macros and 24 MP are more than enough for grainy 135 films. I wonder if a Foveon sensor, set to b/w, would deliver better results but I’m too lazy to find out. In the beginning of some scanning projects I tried to max the quality. These days I’m happy with good enough quality, quickly gained, quickly imported and edited. Some versions of C1 were not suitable for color (or even grayscale) conversions and crashed often, so I used DxO PL and set some styles. Only thing was and still is I could not collect images to projects, albums, groups. C1 got rid of it’s crashes, it’s maybe not the super best RAW converter - but neither my old pictures are the super best grayscale images, so it suits. :grin:

I never tried it. But I think (and as others said in this topic, now it has loaded in and I can see it all :pensive: ) is that DeepPrime will try to fix the noise by your camera.

And the ‘grain’ from the film is ‘detail’ for the camera. So if DeepPrime is doing a good job, it will preserve the detail (film grain) but remove the noise (digital camera capture noise).

Things like Topaz or ON1 have a denoising model to be used on RGB data. But they are still trained for camera noise, not film grain. Had some good results though.

But like others said, most often I add simulated film grain to digital images these days, instead of trying to remove the film grain from my analog film scans :slight_smile: .

A trick - completely off-topic to DxO - that I used for a long time:
Take any denoiser, it may be blunt and crude. If you have nothing, a good blur algorithm will do, I guess. Take your original (inverted?) scan, make a copy of that layer, and then denoise (heavily, really no grain left. Never mind details lost!) the newly created layer.

So now, you have your original scan, and a layer of that same scan but with no grain at all (but probably details lost). You can then create a third layer to compare between the two, and create an overlay layer. So this layer only contains the stuff that is blurred/denoised away from layer 2, in a format that you can set the blending mode to ‘linear light’ and it will show perfectly again.

So blurred-layer (2) with overlay layer (3) blended on top equals original detailed scan (1).
You basically now separated the grain from the scan.

Now do any editing you want on layer (2). Specially, contrast changes will now not affect the noise. Sharpen it if you think it needs it. Resize it to what you want. Increase the saturation… do what you want with it. But make sure you do it on all the layers (the whole document).

Since you’re not editing on the removed grain, it stays much more ‘subtle’ to how it was originally scanned.
When you’re done with the image, you can change the opacity of the grain layer (3) to determine how much of it you want back in, or you even desaturate the grain layer to remove any color-noise in the grain, etc…

So it was a way for me to keep the original scanned film grain, but keep it away and unaffected during editing of the image.

I don’t do enough film scanning these days to bother with it, though. So most images get a bit of contrast / color boost and then just a downscale to target resolution, and I’m done with it.

3 Likes

No Joachim. The film grain does not get affected at all by DP. The best practise to get rid of too coarse grain is to reduce the impact of Microcontrast completele (the slider all the way to the left and then add Fine Contrast istead by 30 to maybe 50). Also be very careful with the Clear View Plus, because it’s using Microcontrast too.

Wow, thank you @jorismak this is a very dedicated way to deal with scans. Sort of frequency separation (what I sometimes use to get rid of some unwanted aperture halos), and I will keep that in mind if I ever find an old film image worth the effort.

Yes, I was thinking along the same way. As DxO profiled the camera already, they also found a recipe to deal with sensor noise. And grayscale grain is no sensor noise. Ocassionally I also reduce clarity

Hi,
maybe a little bit off topic, but I would like to recommend RawTherapee to handle film grain in your scans. You won’t get rid of the grain (and this is not intended in any way when using film, I think), but the scanned grain structure will look more pleasing to the eye in a very subtle and nice way.

Raw Therapee is also helpful to reduce a kind of “colour noise” my Nikon Coolscan produces when scanning colour negatives. An example:


Greetings
Kay

I stumbled upon this thread by accident. It is a thread that was started long, and I have to admit first that I didn’t read all entries.
I’ve used Vuescan many years for slide scans and B&W negative scans, the program is excellent. But the RAW file generated by Vuescan in .dng format is very different from the RAW .dng files from digital cameras. They are mainly used to re-process them later (in Vuescan), if desired, without the need to scan again.
So I’m not surprised that they can’t be opened in PL. I know they can be opened in ACR, but I am aware of several other RAW converters that can’t handle Vuescans .dng’s.
Of course DP and DP-XD are useless, because they are camera-specific. The best way to handle noise in film scans is, in my experience, the use of a dedicated application like Neat Image. N.I. will analyze a homogeneous part of the image, make a frequency-dependent noise profile and there are many parameters that you can tune to eliminate the noise to taste.
This is the above image, processed with more or less default parameters, in Neat Image:

1 Like

Wow… thanks for sharing this. The resulting image looks a little bit “watery” for my taste but in general the outcome of “Neat Image” seems impressive. Could be a game changer for my analogue workflow. I definitely will give it a try. Thanks again!

Yes, you will lose some detail. But mind, that this is a quick-and-dirty example without tuning. You can make settings for luminance noise and chroma noise independently and many more. Expect a learning curve to get to the best possible results. If you want to preserve maximum detail and don’t want a “plastic” look, it is often the best option to keep some luminance noise.
I don’t know the present situation, but when I started to use the application I could download a trial version for free, that had some more simple options and that could save in jpeg only. The “Pro” version can handle 16bits/colour and save in 8 or 16b, jpeg and tiff.
In my experience (and I used it for many slide and negative scans, and before the noise reduction in RAW converters became so sophisticated for digital images as well) it is more than worth the price.

1 Like