A high res 'scan' experiment

[pinholecam]

I've always wanted to try this out, but did not have a chance since I did not have a way to guide the DSLR along a fixed track to do the copy from film to DSLR.
So incidentally, I got a set of 2 macro sliders off ebay for my macro stuff and I jumped on the chance to do this experiment.

There is also this experiment done by another CSer (and coincidentally another Pentaxian )
http://www.clubsnap.com/forums/medi...38-60mp-scan-6x6-mf-film-using-12mp-dslr.html


This is a stitch or 9 frames done using a K5+100mm macro, shot at f8 in RAW.
The image was then converted to JPEG and stitched in Photoshop elements using the Panostitch function.
This was followed by color invert and other typical film to digital adjustments (eg. color correction, level, sharpen)

The original file was a stitch of 9 images using a K5 and 100mm macro.
Original file was 40MB.
8466x7053.

This is a 1024 version

At about 100% crop

Another crop 100%

If you want to see the quality, I have left a larger file here, but too bad flickr only allows me up to 1706 x 2048.
http://farm9.staticflickr.com/8051/8076798288_60780a6ecf_k.jpg

 

[crispy12]

Wow did you use any sharpening at all? Looks good. Lots of detail. It's a nice photo as well.

I stitched a 4x5 large format shot together, file size about 500mb lol. My computer was really struggling.

 

[pinholecam]

Wow did you use any sharpening at all? Looks good. Lots of detail. It's a nice photo as well.

I stitched a 4x5 large format shot together, file size about 500mb lol. My computer was really struggling.

I sharpened it after merge (mentioned in the post)
But it was not OTT.
This is just to see how the whole experiment can get as sort of a 'final product'
That's why I tried to link to a larger file sample for ppl to judge the o/p for themselves.

Personally, I do think this has been the best quality copy I've done so far.
Certainly better than my Canon 9000f.


Wow! 4x5 and 500mb file, I think my PC will crash

 

[hellfire88]

is it possible to see your individual 9 frames before stitching? so as to see how to prepare it for stitch (like how much overlap etc)

no mismatch with the stitch at all? very interesting results, your stitch looks flawless, couldn't tell it went through stitching at all

 

[giantcanopy]

Looks very well done!

 

[Riotbmx]

The details are amazing. Think you forgot to mention the negative used

 

[pinholecam]

is it possible to see your individual 9 frames before stitching? so as to see how to prepare it for stitch (like how much overlap etc)

no mismatch with the stitch at all? very interesting results, your stitch looks flawless, couldn't tell it went through stitching at all

Ok. Let me try to do that over the weekend.
AFAIK, stitch looks ok.
I did somehow screw up the exposure on 1 of the frames though. Maybe 0.3 stop off, but it seems to be 'auto corrected' in the merge.
Each frame was left with 25% or more overlap as is the guideline for auto-stitches.

 

[pinholecam]

Looks very well done!

The details are amazing. Think you forgot to mention the negative used

Thanks.

Film used was Kodak Porta 160.

 

[sunnycamera]

Thanks.

Film used was Kodak Porta 160.

Kodak Porta 160 looks so clean, the fuji 160NS film seems a lot noisy for a ISO 160 film.

The photoshop auto stitch failed me several times, I guess because my photo background is in heavy bokeh, hard for PS to figure out the stitch area.

 

[Riotbmx]

Thanks.

Film used was Kodak Porta 160.

Do you have the final image in colour?

 

[pinholecam]

Kodak Porta 160 looks so clean, the fuji 160NS film seems a lot noisy for a ISO 160 film.

The photoshop auto stitch failed me several times, I guess because my photo background is in heavy bokeh, hard for PS to figure out the stitch area.

Yeah. I'm pretty surprised too.
When down sampled a bit, and then sharpened, I really think I can save money for the time being and not get a FF camera

Initially, I failed using Hugin to stitch.
Luckily, I remembered that Elements has the function too and gave it a try.
My guess is that the algorithm cannot link OOF blur (bokeh) parts, so you would have to make sure each frame has some well defined feature for the algo to relate each frame.

Do you have the final image in colour?

Ok. Color version coming right up

 

[pinholecam]

2048 version is here :
http://farm9.staticflickr.com/8052/8079596502_6899e06d53_k.jpg

I played with a bit of PP this time to get the 'digital sharp' look that we are so accustomed to nowadays.
Resize>NR>sharpen (50%; 0.4pt)

 

[pinholecam]

is it possible to see your individual 9 frames before stitching? so as to see how to prepare it for stitch (like how much overlap etc)

no mismatch with the stitch at all? very interesting results, your stitch looks flawless, couldn't tell it went through stitching at all

Here are the frames :

 

[pinholecam]

 

[pinholecam]

 

[hellfire88]

thanks for showing the individual frames, great insights into the workflow

 

[CLN2011]

Firstly, thank you for sharing. I have a question - how would this compare with direct scanning - say with a V700?

 

[pinholecam]

Firstly, thank you for sharing. I have a question - how would this compare with direct scanning - say with a V700?

I don't have a V700, but I'm using a Canon 9000F.
It beats the 9000F for sure.
The 9000F's true resolution is somewhere around 1500-1600dpi based on online sources.
IIRC, the V700 is at 2400dpi. (or slightly less)

However, I can only advice using the method I've tried as an additional 'bag of tricks' rather than a replacement for any scanner.
Issues abound :
1. WB conversion from Negs>camera>Positive images (not so much a problem with B&W)
2. Potential failure to merge if shots have little to link between frames (eg. if its a shallow DOF shot)
3. Potential failure to merge for god knows what reason the pano software doesn't like


But IMO if it works, it sure is nice