Jump to content

Recommended Posts

10 hours ago, M.Chapman said:

You might want to try the free trial of Topaz Denoise AI. I tried Neat image but  preferred Topaz Denoise AI when pocessing digitised colour slides

 

Mark

 

8 hours ago, MDM said:

Upsizing as a general technique in copying is questionable as you are adding in pixels that were never there in the first place.

 

 

Thanks for your suggestions. I started with the original TIF which was processed from RAW using Photoshop with only minimal image modifications and without the up- and downsizing.

I downloaded the free trial of Topaz Denoise AI The slider was set at 34. I then did the same with Neat Image, letting the software sample and set the amount of denoise to apply. Lastly, I went back to the RAW file and applied noise reduction in ACR, slider set at 17.

 

Most of what I have in my backlog of film images is on transparency film, but at the meantime I'm working on some black and whites. Results on Ektachrome, Fujichrome, Kodachrome, etc., might be different. Also, I've had the Topaz software for all of a few minutes, so any comments have to be taken with several grains of salt. That said, I'm still leaning toward Neat Image as the best balance of noise and detail.

 

http://dondouglas.com/A5_second_test.jpg

Share this post


Link to post
Share on other sites
42 minutes ago, DDoug said:

 

 

Thanks for your suggestions. I started with the original TIF which was processed from RAW using Photoshop with only minimal image modifications and without the up- and downsizing.

I downloaded the free trial of Topaz Denoise AI The slider was set at 34. I then did the same with Neat Image, letting the software sample and set the amount of denoise to apply. Lastly, I went back to the RAW file and applied noise reduction in ACR, slider set at 17.

 

Most of what I have in my backlog of film images is on transparency film, but at the meantime I'm working on some black and whites. Results on Ektachrome, Fujichrome, Kodachrome, etc., might be different. Also, I've had the Topaz software for all of a few minutes, so any comments have to be taken with several grains of salt. That said, I'm still leaning toward Neat Image as the best balance of noise and detail.

 

http://dondouglas.com/A5_second_test.jpg



Film provides some specific challenges - a lot of what we think is noise is probably film grain but is removed by the same methods as noise. I prefer to do all the noise reduction on the raw image,  as it is generally far more effective than working on a TIFF or other converted file format. It rather defeats the objective of getting an optimum balance between sharpening and noise reduction not to work on the raw for all of the noise reduction. That is the main reason I use Lightroom/ACR exclusively rather than any external plugins and I find the noise reduction and sharpening in Lightroom/ACR perfectly adequate. I am not a big fan of Topaz Denoise for the reasons I mentioned before although it does seem to have improved in the latest updates. This gives me the simplest and most efficient workflow, similar to how I work with my normal raw images except I don’t do the two step process of merging two layers in  Photoshop as it is generally unnecessary.

Share this post


Link to post
Share on other sites
40 minutes ago, MDM said:

This gives me the simplest and most efficient workflow, similar to how I work with my normal raw images except I don’t do the two step process of merging two layers in  Photoshop as it is generally unnecessary.

 

I do like the idea of reducing the number of steps and keeping everything from raw processing through keywording in Photoshop would no doubt work best for me. I'll work more with noise and sharpening in ACR. Thanks for the help.

  • Like 1

Share this post


Link to post
Share on other sites
Posted (edited)
25 minutes ago, DDoug said:

 

I do like the idea of reducing the number of steps and keeping everything from raw processing through keywording in Photoshop would no doubt work best for me. I'll work more with noise and sharpening in ACR. Thanks for the help.


Don’t mention it. There is no single  best way to do this.  My approach really was to find a method that would balance decent image quality, sufficient to pass Alamy QC and to make decent prints at A4 size at a minimum, with efficiency of effort. The ES-1 adapter is incredibly convenient and precise in terms of achieving alignment of the film as you are no doubt aware. Your input on your Fuji setup is really interesting in fact as one of the problems in advising people relates to the exact setup.At some point I would like to summarise all of this in an article or video perhaps. The great thing is that we have definitely learned so much from each other on this thread. 

Edited by MDM

Share this post


Link to post
Share on other sites
30 minutes ago, DDoug said:

Thanks for the help

Well thanks for yours, it's a great addition to this thread, particularly as the Nikon ES-1 had already been singled out as the most affordable route into high quality DSLR scanning.

 

I'm not sure if you're uploading any of these 'scans' to Alamy but I note that you tend to downsize them to 4000 px on the longest side. It may be worth pointing out that some on here downsize to the minimum 3000 px  to make sure they get though normal QC, or would you be uploading by the archive route anyway?

Share this post


Link to post
Share on other sites

I have a question about the ES1 and ES2 Nikon adapters.

 

How well do they cope with the thicker types of Agfachrome mounts and slides which are in glass mounts - or do you have to take the slides apart and use a different slide mount?

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, geogphotos said:

I have a question about the ES1 and ES2 Nikon adapters.

 

How well do they cope with the thicker types of Agfachrome mounts and slides which are in glass mounts - or do you have to take the slides apart and use a different slide mount?

I think the ES-1 is better for that. I find the FH-2 strip film holder a little fussy to work with, but have no problem with thick mounts.

 

Although the ES-2 is pricey, I think that for strip film it might be better than bothering with the FH-2 as I have been doing. If trying to digitize film at either end of the strip, the imbalance causes it to tip to the side and that's after getting it correctly inserted in the first place.

Edited by DDoug
add

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, Harry Harrison said:

Well thanks for yours, it's a great addition to this thread, particularly as the Nikon ES-1 had already been singled out as the most affordable route into high quality DSLR scanning.

 

I'm not sure if you're uploading any of these 'scans' to Alamy but I note that you tend to downsize them to 4000 px on the longest side. It may be worth pointing out that some on here downsize to the minimum 3000 px  to make sure they get though normal QC, or would you be uploading by the archive route anyway?

 

Thanks back. So far, I've been cautious and have only uploaded scanned 35mm as archival, and only if they had some historical value. The medium format scans have cleared QC.

Edited by DDoug
  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, Harry Harrison said:

Well thanks for yours, it's a great addition to this thread, particularly as the Nikon ES-1 had already been singled out as the most affordable route into high quality DSLR scanning.

 

I'm not sure if you're uploading any of these 'scans' to Alamy but I note that you tend to downsize them to 4000 px on the longest side. It may be worth pointing out that some on here downsize to the minimum 3000 px  to make sure they get though normal QC, or would you be uploading by the archive route anyway?

 

32 minutes ago, DDoug said:

 

Thanks back. So far, I've been cautious and have only uploaded scanned 35mm as archival, and only if they had some historical value. The medium format scans have cleared QC.

 

If the original slides or negs are sharp then it should be no problem passing QC if the post-processing in relation to sharpness and noise is good. I downsize to 3000 pixels longest edge to make sure but I think they would pass at higher pixel sizes. However, there is probably no real benefit to doing that for film copies. So far nothing has failed QC. 

Share this post


Link to post
Share on other sites

There's another approach I've tried for MF film, but it's a work in progress and I'm not sure it will be successful. I built a simple box to go over the Viltrox LED light with a matte board on top with cut-outs for various film sizes. Unlike a scanner, the camera can be set for f/11 or so if there is a problem with film flatness. In doing that, I found that I could see the individual LEDs in the grid that makes up the light, so I covered it with a 13x18cm acetate film sleeve that is frosted on one side.

My last efforts with this were when I was still trying to make it work with my X-T2. I expect results with the X-A5 will be better but haven't tried it yet. I have some images that were scanned on Nikon LS 9000 and Flextight X5 scanners, so there's something to compare results against._XTC4468.jpg

Share this post


Link to post
Share on other sites
Posted (edited)
3 hours ago, DDoug said:

There's another approach I've tried for MF film

Before the Bowens Illumitran I used a copy stand (Durst M605 enlarger with Siriocam camera adapter in place of the head) and a Hancocks lightbox. This lightbox has the correct daylight tubes but in any case I preferred to use flash so since it was an A4 lightbox I fired two Nikon flashguns down into each end, each set on manual power, 1/8 in fact. Naturally I tested it for evenness of illumination for formats up to 6 x 9cms and it was fine, rather surprisingly in fact. I made card masks for different formats as you did. However I also contrived to allow the medium format holder to slide across so that the full width of the frame, the long side, captured the 56mm width of the film so I could do 6x6 & 6x7 in stages then photomerge in Lightroom. I just wanted to get the most out of the setup, so with your X-T2 you would get a tad under 6000 x 6000 px from 6x6 when cropped, 6000 x 7000 px for 6x7 etc. The Hasselblad X5 would give you around 7000 px across the width of the film as it is 3200 dpi for medium format, though clearly there's more to it than just comparing pixel dimensions.

 

It worked very well, was easy to go to quality photomerged RAW images in Lightroom. I would definitely use this method again for medium format rather than use the Illumitran and have to crop the frame down.

Edited by Harry Harrison
  • Upvote 1

Share this post


Link to post
Share on other sites
4 hours ago, MDM said:

Film provides some specific challenges - a lot of what we think is noise is probably film grain but is removed by the same methods as noise.

 

Reducing just the digital noise is easily done by stacking.

 

wim

 

Share this post


Link to post
Share on other sites
Posted (edited)
32 minutes ago, wiskerke said:

 

Reducing just the digital noise is easily done by stacking.

 

wim

 


I think what we are seeing is film grain rather than noise which is probably a misnomer in my case anyway. I have been shooting at ISO64 on a D810 so there should not be any digital noise. I have read about the technique you are referring to but I think it relies on random noise between different frames whereas grain is not random (for the same slide shot multiple times that is). Happy to hear more about this though. 

Edited by MDM

Share this post


Link to post
Share on other sites

I don't know if it's been noted here yet, but clearly we are not the only people doing this during the pandemic. Equipment and supplies like the ES-1&2 are back-ordered and have been for some time (thankfully, since shortly after I received my ES-1). Some lightboxes in short supply too.

 

I ordered the ES-2 as well, for the ease in digitizing negative filmstrips, but recently cancelled the back-ordered order when I figured I could just temporarily put a filmstrip in a fold-over cardboard slide mount and use the ES-1. I haven't tried this yet, as I haven't yet received said slide mounts, which I had to order from a crafts person on Etsy--slide mounts are scarce as well.

 

Also copying some old 120 transparencies using a lightbox/cheap copy stand set-up.

 

This is almost all for digitizing our large collection of family shots from two generations of two families, but I'm also pulling out archival-worthy shots as I go.

Share this post


Link to post
Share on other sites
Posted (edited)
21 minutes ago, Bill Kuta said:

I don't know if it's been noted here yet, but clearly we are not the only people doing this during the pandemic. Equipment and supplies like the ES-1&2 are back-ordered and have been for some time (thankfully, since shortly after I received my ES-1). Some lightboxes in short supply too.

 

I ordered the ES-2 as well, for the ease in digitizing negative filmstrips, but recently cancelled the back-ordered order when I figured I could just temporarily put a filmstrip in a fold-over cardboard slide mount and use the ES-1. I haven't tried this yet, as I haven't yet received said slide mounts, which I had to order from a crafts person on Etsy--slide mounts are scarce as well.

 

Also copying some old 120 transparencies using a lightbox/cheap copy stand set-up.

 

This is almost all for digitizing our large collection of family shots from two generations of two families, but I'm also pulling out archival-worthy shots as I go.


Not surprising as it is a perfect lockdown activity. This thread was started on January 26th just as we in the west were becoming aware of a potential pandemic coming out of Wuhan. Such a shame the various authorities didn’t get their acts together then and before to contain it and things could have been so different. 
 

Edited by MDM

Share this post


Link to post
Share on other sites
Posted (edited)
5 hours ago, MDM said:

I started with the original TIF which was processed from RAW using Photoshop with only minimal image modifications and without the up- and downsizing.

I downloaded the free trial of Topaz Denoise AI The slider was set at 34.

 

When digitising 35mm Velvia slides, I found I got good results with Topaz Denoise AI as follows;

  • Open RAW in PS/ACR with Sharpening set to 0*; Noise reduction 25, 50, 0; Color Noise Reduction 25, 50, 50.
  • Transfer to PS as 16-bit with no resizing (my camera produces a 4,592 x 3448 image)
  • Duplicate layer
  • Apply Topaz denoise AI (not AI clear) using PS Plugin to top layer - usually with default settings 15, 15 (takes about 30 secs to process)
  • Adjust opacity of top layer to adjust amount of NR**
  • Flatten image
  • Downsize to 3000 on long side ready for submission.

*If a previously sharpened image is processed in Topaz AI the edges will become oversharpened and will show artifacts/halos. It doesn't seem to matter where the sharpen slider is in Topaz I find it significantly oversharpened any edges that are already sharp.

**I use the mulitlayer approach because I find Topaz can sometimes make some areas look too "flat" and "unnatural". Reducing the opacity of the top layer down to 50% (for example) allows some noise to be left in. Also adjusting the opacity of a layer in PS gives immediate visual feedback, whereas Topaz takes a second to update preview image when moving the sliders so is less "interactive".

 

I've never tried Topaz on B/W images. I suspect the "AI" part of the algorithm may vary the noise reduction according to the colour of the subject. For example, I found sand grains on beaches seemed to be left largely unaffected, whilst similar sized film grain in a blue sky was removed. The size of the grain also has an influence. There seems to be a size threshold, below which it removes it as "noise", but above which it retains it as "detail". I found it worked ok for me on my 4592 x 3448 images of a 35mm slide. But performance may be different if the image contains more pixels making the grain apear larger (i.e. more pixels/grain).

 

MDM is right, earlier versions of Topaz Denoise AI had problems around the edges of the frame and that's now been fixed.

Now that I've got a faster computer I may give LR/PS/ACR's NR reduction another go as it will be more "interactive" than it was on a  slower computer

 

Mark

   

Edited by M.Chapman

Share this post


Link to post
Share on other sites
22 minutes ago, MDM said:


I think what we are seeing is film grain rather than noise which is probably a misnomer in my case anyway. I have been shooting at ISO64 on a D810 so there should not be any digital noise. I have read about the technique you are referring to but I think it relies on random noise between different frames whereas grain is not random (for the same slide shot multiple times that is). Happy to hear more about this though. 

 

You are correct about the film grain. However if you want/need to correct the digital noise independently from the film grain, like with a non-Bayer or a low pixel count sensor, stacking is the way to do it.

It could well be that using Topaz AI on the film grain alone gives a good result. I have tried some of those plug-ins like Nik Dfine or Topaz years ago on Velvia scans trying to get rid of pepper grain, some worked with samples as well. They must have improved since then. However Photoshop has improved a lot over time as well and it's currently the only thing I use.

 

How to do stacking: select all layers and transform into smart object.

Layer > Smart Objects > Convert to Smart Object.

Then set stack mode to median (or sometimes mean for different subjects).

Layers > Smart Object > Stack Mode > Median.

 

If you have to align layers first, make sure to remove all outliers before aligning, so all even slightly blurred images have to go.

 

wim

Share this post


Link to post
Share on other sites
5 minutes ago, M.Chapman said:

 

When digitising 35mm Velvia slides, I found I got good results with Topaz Denoise AI as follows;

  • Open RAW in PS/ACR with Sharpening set to 0*; Noise reduction 25, 50, 0; Color Noise Reduction 25, 50, 50.
  • Transfer to PS as 16-bit with no resizing (my camera produces a 4,592 x 3448 image)
  • Duplicate layer
  • Apply Topaz denoise AI (not AI clear) using PS Plugin to top layer - usually with default settings 15, 15 (takes about 30 secs to process)
  • Adjust opacity of top layer to adjust amount of NR**
  • Flatten image
  • Downsize to 3000 on long side ready for submission.

*If a previously sharpened image is processed in Topaz AI the edges will become oversharpened and will show artifacts/halos. It doesn't seem to matter where the sharpen slider is in Topaz I find it significantly oversharpened any edges that are already sharp.

**I use the mulitlayer approach because I find Topaz can sometimes make some areas look too "flat" and "unnatural". Reducing the opacity of the top layer down to 50% (for example) allows some noise to be left in. Also adjusting the opacity of a layer in PS gives immediate visual feedback, whereas Topaz takes a second to update preview image when moving the sliders so is less "interactive".

 

I've never tried Topaz on B/W images. I suspect the "AI" part of the algorithm may vary the noise reduction according to the colour of the subject. For example, I found sand grains on beaches seemed to be left largely unaffected, whilst similar sized film grain in a blue sky was removed. The size of the grain also has an influence. There seems to be a size threshold, below which it removes it as "noise", but above which it retains it as "detail". I found it worked ok for me on my 4592 x 3448 images of a 35mm slide. But performance may be different if the image contains more pixels making the grain apear larger (i.e. more pixels/grain).

 

MDM is right, earlier versions of Topaz Denoise AI had problems around the edges of the frame and that's now been fixed.

Now that I've got a faster computer I may give LR/PS/ACR's NR reduction another go as it will be more "interactive" than it was on a  slower computer

 

Mark

   

 

You have said this before somewhere but can you remind me why are you not using Topaz on the raw file rather than on a converted file? I would need to do new tests to check but I assume the results should be better done in TDN on the raw, saved as a DNG and then processed in ACR before tidying off in PS. 

Share this post


Link to post
Share on other sites
1 minute ago, wiskerke said:

 

You are correct about the film grain. However if you want/need to correct the digital noise independently from the film grain, like with a non-Bayer or a low pixel count sensor, stacking is the way to do it.

It could well be that using Topaz AI on the film grain alone gives a good result. I have tried some of those plug-ins like Nik Dfine or Topaz years ago on Velvia scans trying to get rid of pepper grain, some worked with samples as well. They must have improved since then. However Photoshop has improved a lot over time as well and it's currently the only thing I use.

 

How to do stacking: select all layers and transform into smart object.

Layer > Smart Objects > Convert to Smart Object.

Then set stack mode to median (or sometimes mean for different subjects).

Layers > Smart Object > Stack Mode > Median.

 

If you have to align layers first, make sure to remove all outliers before aligning, so all even slightly blurred images have to go.

 

wim

 

Thanks for the input wim. All useful info. 

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, M.Chapman said:

 

 

 

MDM is right, earlier versions of Topaz Denoise AI had problems around the edges of the frame and that's now been fixed.

Now that I've got a faster computer I may give LR/PS/ACR's NR reduction another go as it will be more "interactive" than it was on a  slower computer

 

 

   

 

I have to take that back. I just reprocessed an image in the latest version that I had done before in Topaz Denoise and am seeing the same artefacts around the edges of the image.

 

While I am sure it is possible to get reasonable results with Topaz Denoise,  I will be continuing to use my LR workflow with two versions (one sharpened and one denoised in LR) of the converted raw image on layers in Photoshop. In areas where detail fine detail is present, I use the sharp layer as the noise or grain will not show up here. Where little or no detail is required (e.g. sky) I use the denoised version as this is also where noise or grain will show up. There is a lot of flexibility in the blending in Photoshop by using different brush and layer opacities. For an image where detail is present throughout I just do one raw conversion and balance the noise reduction with the sharpening. After cleaning and maybe a bit of local contrast enhancement, I downsize the image which further helps with sharpening and grain or noise reduction. 

 

 

 

Edited by MDM

Share this post


Link to post
Share on other sites
Posted (edited)
2 hours ago, MDM said:

You have said this before somewhere but can you remind me why are you not using Topaz on the raw file rather than on a converted file? I would need to do new tests to check but I assume the results should be better done in TDN on the raw, saved as a DNG and then processed in ACR before tidying off in PS. 

 

Because Topaz on RAW messes up image colour and levels, and I found it hard to correct in PS or LR afterwards. As far as I recall, I think you found the same?

 

Mark

Edited by M.Chapman

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, MDM said:

I have to take that back. I just reprocessed an image in the latest version that I had done before in Topaz Denoise and am seeing the same artefacts around the edges of the image

 

Oh, that’s strange. The edge whisps problem definitely disappeared when I updated Topaz (I'm using 2.2.7 the latest). I wonder if image size (pixels) may affect performance? Certainly when I downsized before Topaz (to make it run more quickly) I had more problems. I do recal finding Topaz results a bit unpredictable, but once I settled on a fixed workflow that works for me, performance seems fine.

 

I just tried removing grain from an image in ACR again, and it's a lot easier now I have a faster computer, but I think I still prefer Topaz, but that's probably becuase I'm used to it and have less experience with NR in ACR

 

Mark

Edited by M.Chapman

Share this post


Link to post
Share on other sites
33 minutes ago, M.Chapman said:

 

Oh, that’s strange. The edge whisps problem definitely disappeared when I updated Topaz (I'm using 2.2.7 the latest). I wonder if image size (pixels) may affect performance? Certainly when I downsized before Topaz (to make it run more quickly) I had more problems. I do recal finding Topaz results a bit unpredictable, but once I settled on a fixed workflow that works for me, performance seems fine.

 

I just tried removing grain from an image in ACR again, and it's a lot easier now I have a faster computer, but I think I still prefer Topaz, but that's probably becuase I'm used to it and have less experience with NR in ACR

 

Mark

 

OK mystery solved. I was talking about denoising the raw file in TDN and the artefacts are pretty terrible. When I did it on a TIFF, there were no artefacts. It was also much faster. And yes I recall now about the colour problems with TDN conversions to DNG. I updated to 2.2.7 last night as well with some problems with permissions which was weird. I have not had any problems with other apps. 

 

LR/ACR is definitely the way to go for me as it gives a lot more control over the sharpening and colour denoising as well as the luminance denoising. I will probably check out TDN again on some old portraits at some point.

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, MDM said:

LR/ACR is definitely the way to go for me as it gives a lot more control over the sharpening and colour denoising as well as the luminance denoising.

 

The texture slider seems quite useful for removing film grain.

 

Mark

Edited by M.Chapman

Share this post


Link to post
Share on other sites
Posted (edited)
12 hours ago, M.Chapman said:

 

The texture slider seems quite useful for removing film grain.

 

Mark


I find it good for softening skin as a local adjustment when used judiciously but a bit too coarse for global denoise as it destroys detail. It could be handy though for local denoising. 

Edited by MDM

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.