How many pixels does it take for a digital sensor to outperform 35mm film


Status
Not open for further replies.

bonfire

New Member
Feb 20, 2003
289
0
0
Visit site
#1
We all know the many advantages of digital cameras. But does anyone have a answer to the above question?
 

suhaimig

New Member
Oct 30, 2002
323
0
0
47
Central
#2
bonfire said:
We all know the many advantages of digital cameras. But does anyone have a answer to the above question?
A 5400dpi scan of a 35mm slide or negative results in an image measuring 7800X5232 pixels, or 40+ megapixels. The highest-resolution pro digital SLRs offer about 1/3 of that resolution.

Is it worth to invest a good scanner at the moment? ;)

Cheers. :cool:
 

mpenza

Senior Member
Jan 18, 2002
12,938
0
0
Singapore
www.instagram.com
#3
just wanted to add that having the ability to scan at 5400 dpi does not necessarily mean that film contains that amount of data (i.e. 40MP worth of data).
 

suhaimig

New Member
Oct 30, 2002
323
0
0
47
Central
#4
mpenza said:
just wanted to add that having the ability to scan at 5400 dpi does not necessarily mean that film contains that amount of data (i.e. 40MP worth of data).
You may refer to an article in Photographic magazine, explain what i had said. :cool:
 

Zerstorer

Senior Member
Jul 8, 2002
3,437
0
0
#5
A 4000dpi scanner already resolves Provia down to its grain structure. Scanning any higher results in hardly any real resolution increase. Personally I doubt that 35mm slides contain more than 10MP worth of data, using a 1Ds as a baseline.
 

mpenza

Senior Member
Jan 18, 2002
12,938
0
0
Singapore
www.instagram.com
#6
suhaimig said:
You may refer to an article in Photographic magazine, explain what i had said. :cool:
They may not be right all the time. Not everyone knows sampling theory.

Grain structure, film structure, etc all limits the resolution. Film does not contain infinite resolution that allows more and more details to be digitised with higher resolution scanners. Of course, in the future, some film may allow 40MP of useful data to be extracted.
 

hwchoy

Senior Member
Jul 16, 2003
1,899
0
0
55
Tampines, Singapore.
www.hexazona.com
#7
if the grain size is 1/4000 inch, then you need to sample double that (Venquist theorem I think, I'm going dangerously on ancient memory here, I returned everything to my lecturer 20 years back :) so scanning up to 8000 dpi may still be useful.

However if we're comparing information content, and taking one grain as one pixel, we'd have 4000×36mm×24mm/25.4² (is this the correct dimension of the negative?) = 5670×3780 = 21 Mpix.

However a film grain is not a pixel, the grain structure and packing would contribute to the overall visual quality of the film. I'm not sure if it is meaningful to make such comparisons at all.
 

ST1100

Senior Member
Jun 18, 2003
1,785
0
0
Singapore, Bedok
#9
i've done about 1-2 hundred rolls of scanned film, including negatives, slides and BWs. Then i migrated to a 10D, which has about 6 mpix. My scanner scans at 2700dpi, giving equivalent data to a 8mpix digicam. My favourite print size is A3, and the occasional A4.

If you're not on a pixel count, and concentrating instead on the word 'ourperform', then i confidently say that the 10D decisively and completely thrashes the image quality of scanned film.

i would also add the comment that while you can get about 6000dpi from the highest res BW film, 99% of the time, your system (film, camera, lens, technique) limits your shots to around 2000dpi at best. (And if you habitually shoot at [1/focal length] shutter speed, you get about half that resolution.) As such, this discussion on numbers is purely technical, unless you're a hardcore tripod user.

hwchoy - i don't know if there is a Venquist theorem around, but from your description, it sounds like the Nyquist Theorem. It states that to get data at a certain frequency, you need to sample at at least double the frequency. But like i said, most the data beyond ~2500dpi is purely grain. On normal ISO400 neg film, you see big big blobs of film grain on 2700dpi scans.

Bonfire, if you're really asking about the comparison of image quality between digital and film, i suggest you check out

http://www.normankoren.com

He also discusses in depth other issues besides pixel count: dynamic range, colour/contrast, scanning film, etc. Just skip the overly technical parts. Just comparing pixels does not give a complete picture.
 

suhaimig

New Member
Oct 30, 2002
323
0
0
47
Central
#10
mpenza said:
They may not be right all the time. Not everyone knows sampling theory.
Wow you are better than those man. You should work for the magazine instead. :D

Cheers. :cool:
 

hwchoy

Senior Member
Jul 16, 2003
1,899
0
0
55
Tampines, Singapore.
www.hexazona.com
#11
ST1100 said:
hwchoy - i don't know if there is a Venquist theorem around, but from your description, it sounds like the Nyquist Theorem. It states that to get data at a certain frequency, you need to sample at at least double the frequency. But like i said, most the data beyond ~2500dpi is purely grain. On normal ISO400 neg film, you see big big blobs of film grain on 2700dpi scans.
Yes, it is Nyquist, my bad. Hey still I got 5 characters right out of 7 :bsmilie: not too bad after 20 years :sweat:
 

TME

Senior Member
Jan 19, 2002
6,580
0
0
41
Clementi
#12
ST1100 said:
i've done about 1-2 hundred rolls of scanned film, including negatives, slides and BWs. Then i migrated to a 10D, which has about 6 mpix. My scanner scans at 2700dpi, giving equivalent data to a 8mpix digicam. My favourite print size is A3, and the occasional A4.

If you're not on a pixel count, and concentrating instead on the word 'ourperform', then i confidently say that the 10D decisively and completely thrashes the image quality of scanned film.

i would also add the comment that while you can get about 6000dpi from the highest res BW film, 99% of the time, your system (film, camera, lens, technique) limits your shots to around 2000dpi at best. (And if you habitually shoot at [1/focal length] shutter speed, you get about half that resolution.) As such, this discussion on numbers is purely technical, unless you're a hardcore tripod user.

hwchoy - i don't know if there is a Venquist theorem around, but from your description, it sounds like the Nyquist Theorem. It states that to get data at a certain frequency, you need to sample at at least double the frequency. But like i said, most the data beyond ~2500dpi is purely grain. On normal ISO400 neg film, you see big big blobs of film grain on 2700dpi scans.

Bonfire, if you're really asking about the comparison of image quality between digital and film, i suggest you check out

http://www.normankoren.com

He also discusses in depth other issues besides pixel count: dynamic range, colour/contrast, scanning film, etc. Just skip the overly technical parts. Just comparing pixels does not give a complete picture.
Thanks for the link! Very useful! And I like his landscape shots. Hope I can get something like that (a few percent of his work) when I go to NZ later this year....... :D
 

whoelse

Senior Member
Oct 17, 2003
1,228
0
36
#13
bonfire said:
We all know the many advantages of digital cameras. But does anyone have a answer to the above question?
Throw away spec. take 2 cameras and shoot similar subject on tripod. Print them on paper and see which you like. Result matters not spec. then decided what you like best. If want to compare on screen, you can do also but you are comparing what scanners can produce what result and how this results look like. Meaning how good the scanner is or the scan result is. Nothing to do with the film. :nono:

Forget these opinions, just shoot. Sad to say that there are people (not you all lah) who concern abt all this and at the end of the day we see them produce nothing.

Microsoft says "Where you want to go today?".
I said "What have you produced today?".

:D

"We are part of the world we seek to understand, and our imperfect understanding plays an important role in shaping the events in which we participate."
 

Status
Not open for further replies.
Top Bottom