PDA

View Full Version : Scanner profiles (again)



Giacomo GIRINO
14-Dec-2006, 10:26
Since:
- it8 targets are quite expensive (even those from color aid)
- they are not available for every type of film
- one usually has to rely (i.e. to buy) the profiling software with which the target goes along
- most of all for the sake of indipendence (and showing off: hey!, see what I can do?)
Would it be possible to go this way:
1 - get any reflective target for which the colours are representative (don't need to be known exactely, that, is you don't need a reference file at this time)
2 - shoot a picture of the target on the film to be profiled
3 - by means of a spectrophotometer (some are really affordable: could it be these would be of no use for the current purpose?) measure the image on the film and create a reference file for it
4 - scan the film
5 - write your own routine to create the profile from the reference and scanned files

Possible pitfalls I can think of are:
- a spectrophotometer good enough to this purpose could be too expensive
- although ICC specifications are well documented the icm format is hard to implement
- I have understood absolutely nothing about colour management and profiles

What do you think?

Walter Calahan
14-Dec-2006, 11:18
I think even with a "perfect" profile of everything that is involved with each step, it still takes human interpretation of the image to make it sing.

Chris_Brown
14-Dec-2006, 16:12
Would it be possible to go this way:
1 - get any reflective target for which the colours are representative (don't need to be known exactely, that, is you don't need a reference file at this time)
2 - shoot a picture of the target on the film to be profiled
3 - by means of a spectrophotometer (some are really affordable: could it be these would be of no use for the current purpose?) measure the image on the film and create a reference file for it
4 - scan the film
5 - write your own routine to create the profile from the reference and scanned files

Wow! If you did this, then you could yell from the top Monte Rosa what you've accomplished. Then charge for it!

Step 5 is exactly what profling software does, so why would you write your own? To save money or to make money? ;)

I have a better idea. Find someone with an IT8 target and skilled with the software. You borrow their target and scan it. Give them the file and they make the profile. The cost to make a printer profile in the U.S. ranges from $40 to over $100.

The conditions to this is that you must find someone locally, they must trust you with their IT8 target and you must scan the target correctly (easier said than done).

Good luck, Giacomo! :cool:

Giacomo GIRINO
15-Dec-2006, 02:29
Chris, the purpose of shooting the target and then use the developed image as reference (measure it, write a reference file for it and scan it) would be to have a scan from Velvia that looks/tastes Velvia and a scan from Sensia that looks/tastes Sensia. Could one do that by using a single target for both films which are so different?

joolsb
15-Dec-2006, 02:45
I always thought the point of profiling was to correct for aberrations introduced by the scanner. Therefore, making and using separate profiles for different films is at best pointless and at worst could be counter-productive.

Chris_Brown
15-Dec-2006, 07:24
Chris, the purpose of shooting the target and then use the developed image as reference (measure it, write a reference file for it and scan it) would be to have a scan from Velvia that looks/tastes Velvia and a scan from Sensia that looks/tastes Sensia. Could one do that by using a single target for both films which are so different?

This has been attempted by many who own profiling software, and it's the main method for profling digital cameras. (http://www.gretagmacbeth.com/index/products/products_color-standards/products_colorchecker-charts/products_colorchecker-sg.htm)

However, the two films have radically different gamma characteristics, color pallettes, etc. and I think it would be a software engineering feat if you were to write profiling code that could take two images of a color chart and generate a color profile of the scanner + color neg film.

At least, on this site you'd find a few customers. ;)