35mm print sizes

I’ve just done a quick comparison scan using my Epson V600 on a single 35mm frame (Agfa Vista taken with Kodak Retinette 1B - which is hardly the sharpest tool in the box). The settings used were: Film; Colour Negative, 48-bit colour; and using 5 different resolution levels. Unsharp mask was selected but no other correction or reduction settings were selected.
...

So which did you prefer, and what’s the ‘sweet spot’ for 35mm on your scanner?

I also came to the conclusion that the 3200 scan showed a bit more detail in the twigs on the right of the "middle" tree than either the 2400 or 4800 and above scans. Very interesting test which maybe I should try to replicate with the Plustek!

To recap, the tests at filmscanner.ino suggest that the Plustek as a maximum resolution of 3500 dpi, though you might need to scan at higher resolutions to get that.

The test frame is from a roll of Precisa with some fine detail near the centre of the frame, shot on a Pentax MX last autumn. I didn't remember until after going through all the scans that the lens was the Tokina SZ-X 80-200 zoom. This is a very nice lens for general use, perfect for a day doing autumn colours at the arboretum, but not the sharpest tool to choose for comparisons! Here is the shot:

1611BPMXCS33 by Chris R, on Flickr

I then scanned it at 1200, 1800, 2400, 3600, 4800 and 7200 dpi. The 4800 scan was a “custom” choice in Vuescan Plus; I guess this means it is not automatically offered by the hardware, and may involve some sort of interpolation. Vuescan also offers a pixel reduction option, so I ended up with 3 1800 dpi scans (straight, 3600 reduced by two, and 7200 reduced by 4), 2 2400 dpi scans (straight and 4800 reduced by 2) and 2 3600 dpi scans (straight and 7200 reduced by 2). The files were saved as 16-bit TIFFs, which was a somewhat daft choice as colour depth was not an issue; 8-bit TIFFs would have been fine. Each scan was a single pass, no multi-exposure or dust removal.

It’s worth noting the times and file sizes for the various scan options (not affected by saving at reduced size:

1200 0:18 mins 11.3 MB
1800 0:24 mins 25.2 MB
2400 0:29 mins 44.9 MB
3600 1:10 mins 100.9 MB
4800 3:07 mins 179.4 MB
7200 3:20 mins 403.6 MB

Presumably the 8-bit TIFFs would be half those sizes. By way of comparison, the JPEG of the 2400 dpi scan I made originally when scanning this frame is 1.7 MB (but note, I don't save at particularly high JPEG quality; a Filmdev JPEG scanned at 2000 dpi but saved at high quality is typically 5-6 MB)!

After scanning each frame I took an approximately 800 pixel (across) 100% crop near the centre of each. I have not tried to adjust these crops to match the magnification, but I did look at the lower resolution crops with the loupe at greater than 100% (I couldn’t see an easy way to share the results of the loupe).

The shot was taken in afternoon light with a Pentax MX using (I think) the Tokina SZ-X 80-200 zoom lens at the wider end. https://www.pentaxforums.com/userreviews/tokina-80-200-sz-x-f1-4-5-5-6.html. It is a pretty good zoom lens, especially for its weight, but it does show the problems of trying to draw conclusions from everyday shots like this. I suspect the lens is one of the limitations here, with obvious aberrations at edges.

Since this thread is about print sizes, I also printed all the crops except the 1200 dpi version, side by side (3 rows of 3) on a single sheet of A4 at what would have been approximately 300 dpi for the printed file. I could print to JPEG and include that here but reduced to 1024 pixels across I don’t think it would tell us much!

My conclusions? I didn’t really see any advantage in the 4800 and 7200 dpi versions; the 100% crops of both look very mushy. The 2400 dpi versions look best to me, but the 3600 dpi shots are OK. In the 1800 dpi and 2400 dpi scans I thought the shot reduced from a higher dpi scan was better than the straight scan, but it’s pretty marginal. For the 3600 dpi scan I thought the straight scan was a smidge better than the 7200 dpi scan halved. I normally scan at 2400 dpi with at least two passes and a multi-exposure pass, plus possibly an infra-red pass, but I did not do that comparison this time.

Overall, I’m happy with the way I scan, and happy to go to 3600 dpi for a shot I want to print bigger. Above that, I don’t think I’ll bother.

So, here are the crops:

EDIT: hmmm not working the way I planned, says image too large to upload to TP (800 KB ish). Will need to think again. Check back later!
 

Attachments

  • 1611BCS33 full.jpg
    1611BCS33 full.jpg
    327.1 KB · Views: 14
Last edited:
You are right, might be the upload to TP, I'll try again with flickr...

EDIT: any better?
 
Last edited:
Anyhow, here are the first few scans, from Photobucket this time (!):

a) 1200 dpi:



b) 1800 dpi:



c) 1800 dpi from scanning at 3600 dpi reduced by 2



d) 1800 dpi from scanning at 7200 dpi reduced by 4



e) 2400 dpi:



f) 2400 dpi from scanning at 4800 dpi reduced by 2

 
And some more...

g) 3600 dpi



h) 3600 dpi from scanning at 7200 dpi reduced by 2



i) 4800 dpi (bear in mind this is a "custom setting")



j) 7200 dpi



So, I'm happy to accept this was an inappropriate shot from an inappropriate lens, but having done the work I thought I'd put it up anyway, make of it what you will!
 
You are right, might be the upload to TP, I'll try again with flickr...

EDIT: any better?

Not much change, dunno what others think......and it's just my opinion if you want a basic test shot you need to use a prime on a tripod and sunshine can help to givea a nice contrast (which helps to give the appearance of sharpness esp with a poorer lens). Sunshine like this (double click to enlarge more):-
https://farm6.staticflickr.com/5721/30403237883_12c8a8a768_k.jpg
 
Last edited:
Back
Top