B&W film with Digital Color?

[David Luttmann]

I have in the past taken 3 B&W film photos through R G B filters and combined them in Photoshop to create a color image. I was wondering if anyone has tried this twist:

- Capture one image in B&W with your view camera, and then capture the color image with a DSLR. Strip away the Luminosity channel leaving the A & B color channels to be interpolated to the same size as the film and combine to make the color image.

I'm thinking about a couple of things:

1) Film and processing cost savings
2) Easier processing of color
3) Color channels don't require as high resolution
4) B&W film far outresolves normal color emulsions and as such, the luminosity channel from the B&W scan will provide more detail than a straight scan of color film

Because of the fact that the eye obtains most of it's resolution impression from the luminosity data, the color data can be relatively low in rez and still provide the necessary information.

I'm not talking about using a low end DSLR, but something in the 10 to 15 megapixel range. This should provide good enough color rez and allow greater dynamic range on the color channels.

Any thoughts?

[Bruce Watson]

Interesting idea. I don't see how it can possibly beat an LF capture on negative film. But it might, as you suggest, be sufficient. Hmmm...

My first problem with it is that I don't own a high-end DSLR so the extra cost would be prohibitive. As would the extra weight since I have to carry everything on my back. As would the work to get everything to line up and to get everything to match perspective (should require tight matching of lens focal length equivalents between the cameras which is perhaps not practical).

But if you can get it to work I would love to read about how you did it. Like I said, it's an interesting idea.

[David Luttmann]

I will give it a try. I think the most difficult part will be matching perspective and differences in field distortion between the two. The color channel data should be sufficient. I’m thinking of using my Pentax K20D (15mp) for the color channel data.

I’m not sure how much of a difference this would make on a 16x20…..but I thought it would be interesting to try…..failure or not.

I was just interested to see if anyone shares my bizarre thinking process. My significant other thinks there might be someone……maybe :rolleyes:

[jb7]

Sounds like its worth a shot-
would be interested in your results when you post them-

I suppose its not really critical about focal length, as long as the digital image is slightly wider-
and registration should be easy enough-
just wind the lens back, and replace it with the dslr-
I don't know your camera, but I'm sure it'd be possible with mine-

I suppose its all about interpretation-
its possible that it might be easier to aim for a hand-tinted postcard look,
rather than the full-blown RGB registration exercise you described-

Though, either way, I'd still be interested-

I wonder how lens design would translate-
apart from standard lenses,
most DSLR lenses are telephoto or retrofocus-
and you might be a bit limited with movements-
no more than rise and shift,
without getting into complicated transformations later-
with no real guarantee that you'll be able to match them precisely-

I just did a similar exercise the other day-
(though not for the reasons you describe)
However, the pictures are 100 years apart,
and I didn't have exact registration...

j

[Jiri Vasina]

I think it would be very difficult to match the lenses - focal length, perspective, focus, depth of field, rendition of out-of-focus areas. Then precisely align everything...

IMO, what could be done (a little less difficult) is this - use a low resolution slightly blurred color image to provide the color. I see it as a similarity to the way BW images were hand-colored in past decades.

But anyway, I'd be interested in seeing the results.

[bob carnie]

David
I think if you could solve the registration issues and slightly blur the a and b to help with registration , you may be on to something. I would love to see some results.. since the L channel is supplying all the detail and contrast with a slight colour blur it should be very possible to do what you are considering.
Some of the lens Gurus here should be able to figure out the best possible combinations of lenses ect.
Great idea btw,

Bob

[Peter De Smidt]

I've thought about trying this as well. It's certainly worth a shot.

[David Luttmann]

David
I think if you could solve the registration issues and slightly blur the a and b to help with registration , you may be on to something. I would love to see some results.. since the L channel is supplying all the detail and contrast with a slight colour blur it should be very possible to do what you are considering.
Some of the lens Gurus here should be able to figure out the best possible combinations of lenses ect.
Great idea btw,

Bob

I think I will try using a prime on the DSLR and a close match on the 4x5. I will keep both cameras level to try my best to avoid, as much as possible, distortion between the two. Finally, I'll have to watch the DOF between both....although a slight blurring on the color a & b channels might not be such a bad thing.

I've played a bit with this before, but just in the digital realm. I took a 17mp file from my 1Ds Mk2 and left the L channel intact. I downsampled the A & B channels to 2mp and interpolated them back to 17mp. On a 11x14 print there was absolutely no difference in quality. Just shows how little color plays in the eyes interpretation of detail quality.

I’ll post my results when I’m done with this little challenge.

[bob carnie]

I am slightly blurring the a and b channel and using the opacity slider to taste , sharpening in the L as a matter of course.
not all images like this method but since most of the detail is in the L, the a and b can be blurred without affecting sharpness or detail.
I have spent a lot of time in workshops with Dan Margulis and he is the man when it comes to trying all kinds of effects .

I like your idea and will do some tests and push the a b blur as far as I can, at large magnification to see any adverse effects.