vicmortelmans
Well-known
Hi,
Someone willing to share some b&w postprocessing experience?
My workflow is to shoot mainly b&w film, develop at home and scan the negatives for further processing, web publication and digital printing. I'm fine-tuning my workflow and just published some results here:
http://users.telenet.be/vicmortelmans/fts/processingexperiment/processingexperiment.html
This shows the results for three postprocessing systems:
1) manual postprocessing using the curves tool in the GIMP
2) automatic postprocessing by Vuescan
3) automatic postprocessing by my own algorithm (which I developed mostly out of interest for maths and a little programming)
For illustration, I took three different pictures.
Feel free to have a look and comment on which result you like best.
If you can, please have a go for yourself on the raw image (can be downloaded from the same page), using your own postprocessing tools or techniques, and share your result.
What's also of interest for me, is you idea on the quality of the raw image (which is a combined result from development and scanning parameters).
Groeten,
Vic
PS. if you happen to have some mathematical insight in postprocessing algorithms that are commonly used in tools like Vuescan or other, I'd appreciate to hear about this, so I can further finetune my own postprocessing tool
Someone willing to share some b&w postprocessing experience?
My workflow is to shoot mainly b&w film, develop at home and scan the negatives for further processing, web publication and digital printing. I'm fine-tuning my workflow and just published some results here:
http://users.telenet.be/vicmortelmans/fts/processingexperiment/processingexperiment.html
This shows the results for three postprocessing systems:
1) manual postprocessing using the curves tool in the GIMP
2) automatic postprocessing by Vuescan
3) automatic postprocessing by my own algorithm (which I developed mostly out of interest for maths and a little programming)
For illustration, I took three different pictures.
Feel free to have a look and comment on which result you like best.
If you can, please have a go for yourself on the raw image (can be downloaded from the same page), using your own postprocessing tools or techniques, and share your result.
What's also of interest for me, is you idea on the quality of the raw image (which is a combined result from development and scanning parameters).
Groeten,
Vic
PS. if you happen to have some mathematical insight in postprocessing algorithms that are commonly used in tools like Vuescan or other, I'd appreciate to hear about this, so I can further finetune my own postprocessing tool
Chriscrawfordphoto
Real Men Shoot Film.
I responded to this 5 minutes ago and now my response is gone! Anyway, I downloaded one of your images and worked on it. What I said in my first response was that each image needs its own postprocessing, I have never found an algorithm or standardized processing routine that works at all.
Here's my version:
you can download the full resolution TIFF file with all the layers I used here to see what I did in Photoshop. I think GIMP will open photoshop-created layered TIFFs and let you see the layers.
http://chriscrawfordphoto.com/examples/0112_13.zip
Here's my version:
you can download the full resolution TIFF file with all the layers I used here to see what I did in Photoshop. I think GIMP will open photoshop-created layered TIFFs and let you see the layers.
http://chriscrawfordphoto.com/examples/0112_13.zip
vicmortelmans
Well-known
Hi Chris,
thanks for the feedback. You definitely boosted the contrast! A bit beyond my taste, I must admit. So you're right about there not being any algorithm that fits for all. But they definitely can give a good starting point for most of the shots.
One of the reasonings behind my persuit is that when doing wet printing, you only have two parameters to control the results: contrast and exposure. When going digital (scanning film), I'd like something similar: a very simple postprocessing that allows one or two parameters to be tuned and a result that can keep up with wet prints...
Groeten,
VIc
thanks for the feedback. You definitely boosted the contrast! A bit beyond my taste, I must admit. So you're right about there not being any algorithm that fits for all. But they definitely can give a good starting point for most of the shots.
One of the reasonings behind my persuit is that when doing wet printing, you only have two parameters to control the results: contrast and exposure. When going digital (scanning film), I'd like something similar: a very simple postprocessing that allows one or two parameters to be tuned and a result that can keep up with wet prints...
Groeten,
VIc
charjohncarter
Veteran
vicmotelmans, I used your leaf image. First I used GIMP to invert, light curves, levels. Then I went to LightZone and by isolating zones I lightened only the light leaves. The first image is Gimp, and second is Gimp and LightZone. The third is the LightZone image with unsharp mask 20-50-0
Attachments
Last edited:
vicmortelmans
Well-known
charjohncarter said:
Nice! I like these! Even if it's again a contrast boost. What I also notice is that you got rid of the ennoying grain in the shadow area's. It looks like my shadows are much too bright... blindfolded by the quest for 'shadow detail', I guess, while shadows mostly produce grain, even on this slightly overexposed film.
THanks!
VIc
charjohncarter
Veteran
vicmortelmans, I know, I didn't like that grain so I made it dark. LightZone has the ability to 'freeze' the image and let you work in the area you want. With this one I lightened the foreground leaves. I like that image by the way. I am also on the quest for shadow detail. So, it hurts to kill it at times.
What is that 'Log Pixel' graph?
What is that 'Log Pixel' graph?
Last edited:
MartinP
Veteran
It's the middle of the night so I'm not going to try it now, but us Linux users have both the Gimp and Lightzone available on legal no-cost licencing. Happy, happy, joy, joy (sorry, I should already have gone to sleep).
vicmortelmans
Well-known
charjohncarter said:
Hi,
the log pixel chart shows on the horizontal axis the logarithm (P) of the scanned pixel value and on the vertical axis the logarithm (Q) of the pixel value as a result of post-processing.
I must admit that the graph brings little added value, but I implemented it, because my assumption from a theoretical point of view was :
- the film density steps (*) are linear to the film exposure stops (at least in the center of the so-called 'characteristic curve' of the film)
- the light transmitted by the negative is logarithmic to the density
- the scanned pixel value is linear to the transmitted light
- thus: the steps in the logarithm of the scanned pixel value are linear to the film exposure steps
I imagined a post-processing algorithm that would have as a primary goal the reproduction of the subject's light values on the display. Therefore, it must produce pixel value steps that are linear to the film exposure steps. Or---because of the reasoning above---the algorithm would be basically a linear function between the log P of the scanned pixel values and the log Q of the displayed pixel values.
Just one other thing has to be taken into account: the actual pixel values stored in a bitmap are not linear to the display brightness, but are compensated for the display gamma. This means that the postprocessing algorithm should perform a gamma compensation on the theoretical display pixel values, before storing them in a file.
But by practicing, I notice that the optimal postprocessing is not showing a linear relationship between the logarithms of scanned pixel values and (theoretical) display values, but rather a linear relationship between the plain scanned pixel values and the (theoretical) display values.
This posting was partly to figure out how this works and one of the tests I'll do is to make a wet print of one of my pictures, scan it on a flatbed and compare that result to the digital processing.
Groeten,
Vic
(*) a step or stop being defined as a fixed increment (e.g. +1) in log-space that reflects a fixed product factor in linear space (e.g. x2).
charjohncarter
Veteran
Thanks Vic,
I copied that 'log pixel' essay. I'm afraid I'll have to read it a few times. Logarithms are in my distant past. So far back that we did them on a slide rule.
Carter
I copied that 'log pixel' essay. I'm afraid I'll have to read it a few times. Logarithms are in my distant past. So far back that we did them on a slide rule.
Carter
charjohncarter
Veteran
I'm still reading your essay and it is starting to sink in, thanks again. Please, post your results from your wet scanned print or if you don't want to do that PM me. I would be very interested.
Carter
Carter
vicmortelmans
Well-known
Hi, sorry I didn't post any news... Lack of time causes my project to be put on hold. I'll pick it up again sooner or later and let you know. Thanks for your contributions so far!!!
Groeten,
Vic
Groeten,
Vic