mijosc, does step 6 introduce loss or errors, in your view? The workflow seems like a good one. I am curious to hear about your experience on downsampling.
mijosc, does step 6 introduce loss or errors, in your view? The workflow seems like a good one. I am curious to hear about your experience on downsampling.
In my opinion, no, it does not introduce errors. I do it to size the image to the “real” resolution of the scanner and to reduce scanner noise in deep shadows. I’ve played around with different methods of downsampling (nearest neighbor, bicubic sharper, and bicubic smoother.). The differences are subtle, but noticeable. I’ve settled on bicubic smoother, for the moment; I think it minimizes grain.
Of course you could just scan at 2400 or 1600 dpi and get an image that’s 98% there. But where’s the fun it that!
Originally Posted by chassis
I agree with Mijosc. I usually use bicubric smoother for reduction, and then sharpen as needed at the very end, and depending on use, final size....
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
I thought your summary was excellent. I use a slight variation on your approach (I am not scanning slides either BTW), but I get improved results, i.e. it looks better and it has a sound technical basis
Remember you can't actually change the sample size of these scanners other than switch lenses, (which is what happens when you go from the full size to the transparency area).
If I scan at the highest resolution of the V850, what size file does that give me for 4x5? at 4k it is 1.8gb so anything larger than 4gb and you can't write out a tiff or dng file. I do 16 passes to minimize the noise and currently at 3600 dpi it is a file between 1.37 and 1gb depending on how I crop the slide/negative or if I convert to 16bit gray for final output after scanning at 48bit color, this file is about 450-500mb.
I assign it AdobeRGB1998 in photoshop since it initially comes with no embedded profile. So far, that seems to work. I would also apply the icc profile outside of the scanner only because I am doing raw scans. I would want to compare by applying in the scanner and outside of the scanner to see which is better if either.
You apply the icc profile and then covert to adobe98 or ektascan.
“You often feel tired, not because you've done too much, but because you've done too little of what sparks a light in you.”
― Alexander Den Heijer, Nothing You Don't Already Know
Then if I bring it to camera raw or lightroom, I apply the icc profile first, then assign the color space? I suppose that would be the same as applying it at the scan stage. I would only do this if creating a file other than a raw file. Colorperfect which I am getting familiar with says not to use an icc profile on the raw image, but to just assign the color space in ps prior to bringing into colorperfect. I know for my digital stuff, I apply the icc profile in lightroom at the beginning before any adjustments. Same with camera raw and work on them in a particular color space. When I output the files then I assign the adobergb1998 or srgb, etc embedding that in the photo.
What your, effectively doing here is saying that the raw data from you scanner matches AdobeRGB1998 which is won't but at the same time its not that far away to be a disaster.
So in my example if you had the raw value RGB 65535,0,0 you could now plot that exactly in XYZ space or any equivalent and transform it from there to anywhere in other colorspace. LR AFAIK will use Profoto as its internal model for example.
When you output a standard sRGB jpeg it will transform RGB 65535,0,0 to the appropriate sRGB value. Which BTW will probably result in a clipped value.
Not saying what your doing doesn't work, just trying to be of some assistance.
Thank you for the help. I never say I know it all. The day I do, is the time to quit. I always want to learn a better way or the correct way, etc. Being self-taught leaves a lot of room for improvement. I have produced a lot of horrid images that when I go back after learning something they turn out much better.
I can see what you are saying. By applying the icc profile first, then the colorspace, the color values will be more correct than color space first then icc profile. For lightroom, I believe you have a choice of the working space (not sure and will check this tonight)
Posting Permissions
Reply With Quote
Bookmarks