The B&W scanner bake-off challenge ...

[dmr]

I'm starting a new thread so I can post a poll with it, as JLW suggested in the old thread:

Since this seems to be a topic on which feelings run high, you might want to post the scans WITHOUT indicating which was done with which method.

There are two scans below of the same negative. One was scanned as a B&W negative, the other as a color positive, inverted, converted to greyscale, you know the drill.

Levels were adjusted on both to bring the white and black points to the edge of the histogram. Grey point was adjusted so that both appeared on the same approximate point on the histogram curves, which looked similar, but not identical.

Please tell us which you think is which, comment if you want to tell us why you think so ...

 

[ed1k]

My guess is
dwma.jpg (left) is scanned as b&w
dwmb.jpg (right) is scanned as colour positive
Could you also post small crops of each image with 100% details (without downsizing but just the same small part of the two images with resolution as you scanned it).
Eduard.

 

[dmr]

Could you also post small crops of each image with 100% details (without downsizing but just the same small part of the two images with resolution as you scanned it).

These crops are from the original scans, the left one being from the same original scan as the left above and vice versa. These are 3200dpi and yes, you can see the detail of the film grain quite clearly ...

Hint: If you have a browser with tabs, open each of these in a new tab and flip back and forth.

 

[jan normandale]

well dmr... and just how are you this fine morning? I'm looking at these shots and scans and I have to say it is difficult because of monitor resolution issues. Looking at the first set it is difficult to get past the question "is there a significant gain"?

Looking at the enlarged crop in your second post it becomes evident that the second crop has less 'grain' than the first crop. So this forces a second question. Do I as a person who looks at photographs look for image quality or image (ie science vs art)? I'm a category two kind of person so image quality is only a secondary consideration after the image itself.

Net conclusion.. sometimes I want image quality especially if I am making a large print. Other times I deliberately want a grainy blown out look to a print. It all depends. I'm sure after cruising thru your galleries that you are also a category two kind of person.

What did I just say? I'm not sure. But I'd like to know eventually the technique you used for the second jpg. But let this run for a while. I'd like to read what others have to say and think. Nice thread.

Jan

BTW "bake off" ??? Kind of freudian n'est ce pas? My favourite is apple flan glazed with apricot... yours?? Heh!

 

[dmr]

Looking at the enlarged crop in your second post it becomes evident that the second crop has less 'grain' than the first crop.

This was really the most eye-opening revelation of this whole thing. Up until then I was mostly looking at 2 images side by side in Photoshop's "tile" mode. Bringing up the full size crops really showed the difference.

Note that the grain structure is clearly visible in both, but much different in appearance.

BTW "bake off" ??? Kind of freudian n'est ce pas? My favourite is apple flan glazed with apricot... yours?? Heh!

I don't know why I used "bake off" for the title, nothing in particular. I really don't bake much of anything anymore and when I did is was usually stuff like a cake from a mix. I don't ever remember entering in a real bake-off. 🙂 I've done from-scratch on occasion, but that's a lot of work. 🙂

 

[ed1k]

Looking at the enlarged crop in your second post it becomes evident that the second crop has less 'grain' than the first crop.

I think the grain is the same, but it less 'visible' on second (b) crop because of tonality. Well, I do not change my mind (even if it turns out to be wrong). My theory is it's dificult for scanner to record information in dense parts (highlihts in positive); these dense parts of negative is the most noisy part of image. For negatives it's not that bad, becase converted to positives these dense parts become highlights where noise is not very visible. It looks like I see more details in highlights on second (right) image; but that could be noise or left image got highlights cutted off while tweaking with histogramm... Scanning as RGB is not exactly 3x multiscan but I assume you get three channels which may be correlated to some degree but resulting grey component should be more accurate. To reduce noise one needs multiscan with different exposure and smart postprocessing - I doubt "multiscan" mode of KM software does different exposures, but some averaging also could help. I found scanning as RGB is faster than scanning with 2x, so that's what I do. During my tests I was concentrated on noise in dark sides of negatives (as I like clean highlights, not bleached out), maybe I missed something important.
Eduard.

 

[aad]

Hard to say-differences between scanners makes more difference than seen here. Is it OK to say it doesn't matter, that one should do what one likes?

 

[pphuang]

When scanning as a color positive, did you scan as an 8 or 16 bit file? I always wondered if there might be an advantage to scanning a BW negative as a 16 bit color positive in terms of editing, perhaps allowing you to capture and retain more of the tonal range in the negative, but I've never looked at this in any systematic fashion - too lazy, I guess 🙂

 

[kipkeston]

I'd say the grains look different. the one ending in 32 looks more like color grain. Who knows.

 

[Bob Michaels]

apologies for not looking at your 'bake off" JPG's, but I done enough of my own (Minolta MultiPro using Vuescan) to know that I cannot tell the difference in a file scanned as Greyscale or RGB, except in the size of the file.

 

[MikeL]

I guessed the left one was scanned as a positive and inverted due to it looking a bit flatter. Whenver I've done this the image needs curves work even after adjusting the black and white range.

 

[rich815]

I've tried like heck to see any differences using these two methods with my Epson flatbeds, my Nikon 4000 and Nikon 9000 and have never found any benefit.

 

[ed1k]

I scan b&w as colour positive 16 bit readout every time and do post processing in PS. Yes, the curve need to be adjusted. So basically I do conversion to grayscale, then inversion (ctrl-I), curves(ctrl-M) and levels (ctrl-L). Maybe some good software may do that job reasonable well, especially if you scan a lot and don't want to spend much time.

Edit: By I don't see any answers from the author. Is the poll correct answering original question?

 

[Doug]

Left scans have more shadow detail, so I'll guess that's the inverted positive scan.

 

[amoz]

Left scans have more shadow detail, so I'll guess that's the inverted positive scan.

I tend to follow the same reasoning. What's the correct answer?

 

[dmr]

I tend to follow the same reasoning. What's the correct answer?

LOL! This thread is almost 2 years over. I'm going to have to find the originals on the old PC to be sure! 🙂

 

[amoz]

LOL! This thread is almost 2 years over. I'm going to have to find the originals on the old PC to be sure! 🙂

😀 I know but I never noticed this thread, until it popped up again yesterday. Good luck digging up the data on a dumped PC... that would take another two years for me 😎

 

[jja]

I agree w/ Eduard. In my experience (using an SD iv, like yours?), scanning in color makes for more 'muddy' (darker, blocked, whatever) scans than scanning in b&w. On the other hand, scanning in b&w, tones sometimes get washed out. Assuming you did not do a lot of equalizing in photoshop, I'd say the left is a little lighter, "better" (subjective) shadow detail.

Edit: I see that there was some equalization, but I'll stick w/ my choices.