digital vs film - again with attached crops

[neil l]

I know this has been done to death but I thought you might be

interested in seeing these crops I have done today whilst preparing

some shots for stock. I shot my D70 alongside a Bronica SQA mainly as

security in case the film spoiled in the xrays. <p>

 

The film was canned on an epson 4990, and I up-rezzed the d70 shot

with bicubic smoother to the same pixel size them did the crops

attached maybe I scanned it wrongly but to me the D70 looks to have

more detail which of course cannot be right. <p>

 

I know which one I prefer! <p>

 

P.S I hope I can attach the files OK otherwise this message will make

no sense at all!!

[._kaa]

Looks like you failed with file attachments... :-)

[neil l]

BTW the crops can be found in my gallery, see if you can tell the D70 Vs the Bronica.

[._kaa]

You understand that you are not comparing Bronica vs D70, you are comparing your scanner vs D70, right..?

[marbing]

Well, you made a couple of methodological errors in your comparison but, don't feel bad, these mistakes have been made for years by most digital people trying to prove their point.

 

1) As pointed out, you are NOT comparing your D70 to your Bronica SQA! You are comparing the digital output of your camera to the digital output of your scanner.

 

2) Did you scan directly from film or from a print? Every step from the original loses information so that a copy of a copy of a copy of an original is never as good as the original. You are comparing a first generation digital image to a second (or third) generation copy of the original film image.

 

3) You artificially enhanced the digital image by using editing tools to boost resolution. So, now you are comparing a degraded copy of a film image to an enhanced digital image. Are we still on a 'level playing field'? No, so the results can't be trusted.

 

4)Did you take some of your 'crops' from the edges of the image? If so, you have run into differences with DoF between digital and film cameras...digital has a much longer DoF and everything appears to be in focus while film cameras can render background objects softer.

 

How about this? We make prints from both cameras (regular lab enlargements for the film camera and regular digital prints from the digital camera) and THEN we scan BOTH sets of prints. I think you would see significantly different results. Both images would then be 'copys' of their respective 'originals'. The 'scan' is a copy of the copy...but both cameras are at the same level. Of course, the scan is only necessary for posting to the web (which is a digital medium). You can skip the scan and compare the two prints with a loupe and you will see how the digital shot disintegrates into unrecognisable blobs while objects are still identifiable on the film print.

 

As you can see, I don't have much time for these so-called comparison shots because they are usually manipulated to prove the point the tester wants to prove and not representative of the relative merits of each system.

[40mm]

What Meryl said, but also you seem to have not provided any detail information for the pictures in your folder, so I can't know which picture is from film and which from digital, anyhow.

[yuri_wolf]

Meryl Arbing: <I>You artificially enhanced the digital image by using editing tools to boost resolution.</I><BR><BR>

 

Please, share the knowlege - what editing tools can be used to boost <B>resolution</B> (as opposed to sharpness, contrast, number of pixels etc.)? I would love to see that old codger Claude Shannon finally proven wrong! :-)

[neal_shields]

As usual this is NOT "digital vs Film" it is digital vs scanner.

 

Film tests out to resolve 80 to 200 Lp/mm

 

Why do people persist in thinking that they can take a scanner that isn't capable of 1/5 that resolution and test film with it.

 

If you don't believe me, maybe Zeiss?

 

http://www.zeiss.de/C12567A8003B58B9/Inhalt-Frame/C848A011BACB8FF0C1256B1A002DBD21

 

 

Or put you camera on a tripod, focus carefully and release the shutter with a cable release or self timer and then go get a drum scan done at 12000 dpi.

 

You will find as I did, that fill captures way more information than digital cameras, (excepting of course Better Light backs.)

 

Do people think that ALL the film companies and lens makers have being faking their resolution numbers for the last 50 years???

 

Was this a conspiracy against digital cameras before digital cameras were invented???

 

You have a very fine camera that meets your needs and most people?s (especially hand held), be happy with it.

[marbing]

I understand that products like Genuine Fractals increase image resolution by interpolating pixels that do not exist in the original image and so, increase pixel count. It creates false 'detail' using a fractal algorithm. This is quite transparent in 'natural' settings (such as leaves on trees etc.) but is less successful with straight verticals and horizontals. Of course, that is only one example.

[neil l]

Hello all <p>

 

Quite a response! <p>

 

I like most people cannot afford or justify a drum scan for every shot I send into a library and something like the 4990 is the only reallistic option I have. <p>

 

For the record I interpolated this with PS bicubic not Genuine Fractals and the crop is from the upper middle part of the frame on one and dead centre on the other. File names crop 2 and 3 are the scanned provia. <p>

 

I don't think there is any doubt at all that a standard print from the D70 wouldn't stand up to a ciba print from the tranny. But the point I am making is that with the right tools (Genuine Fractals would no doubt have retuned a better result still) you can get something from a 6MP DSLR that holds up to scanned 6x6 film. No doubt better results could be gleaned from the 4990 but the fact that this was even close has made me re-evaluate my views. WE are not just talking about someone saying "My 18x12 prints from my DSLR look great" This is actual detail when zoomed in 200% and cropped<p>

 

I know this is my D70 vs my scanner but isn't it all about the end result? Stock libraries don't take slides or prints they take digital files. <p>

[daniel_taylor]

"I know this is my D70 vs my scanner but isn't it all about the end result?"

 

Yes, it is, but this point is lost on most "film defenders". If you own an Imacon or are willing to pay for drum scans then the MF will pull ahead at larger print sizes. If you own a flatbed then it's valid to ask whether or not it's worth dragging out the MF equipment when the smaller, lighter D70 will produce much better prints. Some people constantly criticize the digital workflow, but the best film workflows are difficult and expensive, and anything less compromises the original image.

 

Your Bronica is something to use when you know you're going to pay for a drum scan. For everything else, it's pretty obvious which camera to use....

[gary_woodard]

"Stock agencies dont take slides", news to me, which stocks agencys wont

accept slides may I ask.�

[loreneidahl]

how about comparing your d70 or my D2h to my 4x5 negatives using my 4x5 enlarger to create the final output or my 6x7 negative enlarged using a 6x7 enlarger to crete the final output?

 

This would eliminate a film to scanner conversion. That would keep digital = digital and film = film.

 

 

 

 

just pick a camera - any camera, and shoot.

 

This is old stuff.

[neil l]

It is a dilema and I know has probably been discussed to death. On one hand to my (untrained) eye an interpolated digital image seems to hold up well to scanned MF at least on my 4990. However most stock libraries are still reluctant to take digital files (alamy will provided they are interpolated with GF or similar). <p>

 

is a minefield I agree but without access to a high quality scanner I am starting to feel I am one piece short in my jigsaw. If it wasn't for my desire to grow my stock collection I'd ditch MF and stick with digital, it is easier and cheaper and great for my wedding work (once you have the equipment that is) but despite what my eyes tell me nothing below the 1DS mkII has any credibility with the top libraries right now. <p>

 

Here's another question would 35mm scanned on a high quality film scanner (ie. minolta 5400 level) give better results than MF scanned on my 4990??.

[thomas_sullivan]

"...would 35mm scanned on a high quality film scanner (ie. minolta 5400 level) give better results than MF scanned on my 4990??...."

 

dont know...........but your MF scans would give better results if you scanned them on a high quality medium format film scanner (Nikon 9000, etc), that's an absolute.........and that is the test that should have been made here. That would also take into account the "at home consumer" level ability to do those scans.

 

would your Bronica then beat out the D70...........who cares. It would be the best you could do at home for a reasonable price. Frankly, I think they would probably pretty much a match....or close enough to a be a "who cares".

[neil l]

This is getting interesting!! and for me illustrates that there is a lot more to consider than simply pixels. <p>

 

Firstly it would appear that I have over-rated the Epson 4990, along with the 4870 it has recieved great reviews with a concensus that it is not great for 35mm but very good on medium and large format. It seems that everything is relative and whilst my results have varied (crops 1 and 2 being the worst example, 3 & 4 being a little closer) for me I will say that.....drum roll.....<p>

 

<b>Files from my D70 once interpolated are at least the equal and I think better than 6x6 scans from my SQA and Epson 4990. </b> <p>

 

This is quite a statement I know but a very relevant I think. I'm guessing there are pros that use the 4990 (or canon equivalent) and like me there must be a lot of people who shoot stock who simply cannot afford an Imacon or justify the cost of drum scans. The questions I would than ask compared to my current set up are: <p>

 

If any of you guys shoot MF for stock how are you scanning? and what sort of feedback are you getting from the image library's? <p>

 

Would a 35mm system with a better quality scanner return higher quality files? <p>

 

Is it worth selling up the MF gear and going 100% digital? (another thread this one probably) <p>

 

All this discovery has done is got me thinking, I thought the day I went 100% digital was at least 2-3 years away, I am now thinking it could be quite soon! If a smaller lighter Nikon body comes out with the 12.4 MP sensor then who knows that could be it! and the S3 pro is all of a sudden looking quite tasty!.<p>

 

The fact is that whatever my eyes tell me Alamy along with other libraries would still rather take the 6x6 scan from my 4990 than an interpolated file from my D70 no matter how they look and that may remain the biggest factor of all.

[yuri_wolf]

<I>...increase image resolution by interpolating pixels that do not exist in the original image and so, increase pixel count. It creates false 'detail'...</I><BR><BR>

 

"Pixel count" - yes. "False detail" - yes. True resolution - I wouldn't say so. You can't re-create the information that hasn't been captured originally - only fake it (thus the reference to Shannon in my first post in this thread).<BR><BR>

 

All of this would have been usless pedantic nitpicking but let's return to the start of the discussion. Upsampling the digital capture to the pixel count of the scan <B>is</B> a reasonably fair comparison, if only slightly loaded against digital<SUP>*</SUP>. You do not "artifically boost resolution" this way, which is simply impossible, you are adjusting the precieved image size to make visual comparison easier.<BR><BR>

 

*- Why? Because you when you upsample the smaller image you magnify all the imperfection and artifacts that existed there, comparing them with "original" capture - the scan in NL's case. If you downsample the larger image to the size of the smaller one, you potentially throw away the details that were there, so if you want to load your test in favour of digital capture versus a high-res film scan, this is the way to go. :-)

[scott_eaton]

<I>Do people think that ALL the film companies and lens makers have being faking their resolution numbers for the last 50 years???</i><P>No, just publishing those <b>inflated</b> numbers using <b>anechoic</b> room tests based on high contrast, <b>monochrome</b> test targets at the film's <b>ideal</b> ISO. I don't shoot high contrast, monochrome test targets for a living, and when it comes to <b>real world</b> situations involving mixed contrast and color gradient ranges, conventional dye based color films, even the slow speeds ones, fall off to a fraction the inflated specs you keep posting those regurgitated hyperlinks to. <I><P>You will find as I did, that fill captures way more information than digital cameras,</i><P>Can you provide us with some examples?<I><P>how about comparing your d70 or my D2h to my 4x5 negatives using my 4x5 enlarger to create the final output or my 6x7 negative enlarged using a 6x7 enlarger to crete the final output?</i><P>Gee, what happened to comparing 6mp dSLR capture to 35mm film? Now all of a sudden it's a D70 vs MF and LF when I thought Kodachrome 64 was the equivelant to 20megapixels? {smirk}.<P>To cut to the chase on this, I don't think anybody here is going to rave about the resolving power of an epson 4990, so the question is not 'scanner vs film' but how much scan quality is sufficient to beat a 6mp dSLR.<P>Most of the latest desktop scanners I see posted samples from on photo.net show frikken molecules in the emulsion at 4000dpi and 100 speed slide film. In this respect I'm laughing at those here claiming an optical print from film is going to show more resolution than a proper scan. Not at 4000dpi, and not if the scan is done correct - sorry.<P>I occasionally shoot metal parts at work for quality control purposes, and use my 10D with 100mm2.8 macro, and my RB67 with extension tubes. With a 2000 or 4000dpi scans from a Nikon MF scanner my 6x7 trannies show far greater detail and sharpness than the 10D capture. I've tried the same tst with 35mm, and 100 speed Provia scanned at 4000dpi showed a bit more detail in low contrast areas than my 10D capture, but the 10D images were cleaner and more neutral and the sharpness difference was not obvious once the 10D images were adjusted for Canon's soft CMOS capture. <P>All in all a 2000dpi scan from low speed MF transparency film should 'spank' 6mp dSLR capture in terms of sharpness and raw resolution *if* the scan is of good quality.

[steve_swinehart]

"... it's worth dragging out the MF equipment when the smaller, lighter D70 will produce much better prints."

 

I'll bet my Plaubel Makina is lighter, and in most dimensions smaller than your D70. Actually, I'd almost put my 500CM with the 50mm up against the D70 as it's really not all that much larger than my N-90 with the 35-70 zoom.

 

The problem with this argument of digital versus medium format or even film in general is the equipment used (either film or digital), and the desired end result.

 

If you're only going to do 11x14 or smaller, the D70 is probably an adequate choice. However, once you get to 13x19, or in my case much larger, only a medium or large format film camera really works, unless you have the reason or resources to purchase or lease a medium or large format digital back.

 

The film grain argument is a bit of a canard. Film grain, like digital noise can be greatly reduced by using a noise reduction program like Neat Image. If it's fair to use this type of processing on a digitally created image, then it should be just as workable and fair to use on a scanned film image.

 

Comparing a digital image to an unprocessed (noise reduced) film image, is in my estimation, not a fair comparison as you need to treat BOTH images to the best processing possible and THEN compare the results. Any other way is loading the comparison in favor of one technology.

 

I make 22x27-inch and larger prints from 6x7. They are absolutely grainless at that size and will hold up to "nose on the image" scrutiny for detail - I know, that's how I evaluate them.

 

I have never been a big fan of "miniature cameras" (35mm). I do use them, but find that 13x19 is really the maximum where they'll hold up to the nose on the print test. After that (16x20), unless the film is low speed, and the exposure absolutely perfect, the image starts falling apart to the point that I don't like the end result.

 

On the flip side, I read posts by people who claim they make 20x24's from 6 megapixel cameras and "they look great." I can only assume their defintion of "great" is somewhat different than mine, as my 6 megapixel camera RAW images look like horse pucky at anything larger than 8x10.

 

I've done test prints for a friend who's a Leica rep from his D2, and he agrees with me, that no amount of interpolation makes a "good" image larger than 8x10. Workable yes. Presentable at the correct viewing distance .... hmmmmmmm....marginally. As good as 35mm at the larger size? Not to me.

 

DSLR's are a different story as the imagers are larger and can stand more enlargment. But, even the 14-16 mp cameras don't look really good at 20x24 unless the subject is correct.

 

That's the other part of the digital conundrum. Some subjects will look good at larger sizes. Landscapes are the worst, portraits are probably the best subjects for big enlargments. My point is that it's so interactive that making blanket judgements and statements are really just generalities and not absolute facts.

 

I use relatively high end equipment. When I see comparisons based on low priced flatbeds, it's a joke in one way, but makes absolute sense in another. You can't get the most out of the film from a flatbed - not even close. I own an Imacon, and won't put the results up againts a drum scanner - but, it's the best I can justify personally.

 

Isn't that what your really testing? What you can justify personally for film versus what you can justify personally in digital?

 

If the choice is buying an Imacon or buying a D2x or even a Canon 1DS Mark II - and you never want to make a print larger than a certain size, or of large finely detailed landscapes - the digital camera wins.

 

If you own 35mm to 4x5 cameras and want to make prints from all of them - especially large prints - the same amount of money is better spent on an Imacon.

 

There are no absolutes in this "demonstration" despite the unequivocal statement "I know which one I prefer!"

 

Then I'd have to ask, which one would you prefer at 20x24 if the film is scanned and handled properly? Probably NOT the D70...

 

Use, versus desired results, versus budget. That's what it's really about.

 

 

[neil l]

I think maybe Scott has missed my point in his defence of the good ship film. I am in no way saying that my D70 has more detail than a 6x6 slide from my SQA. I am saying that to my eye images from my D70 (when interpolated in PS) are a match for a 6x6 scan from my 4990. If I was to use Genuine Fractals then maybe the digitals would look better still?? <p>

 

I have absolutely no doubt whatsoever that if I had a decent film scanner it would be different and if I had an imacon or drum scanner it would be no contest, but I don't. I have a 4990 and have to work with the tools availanble. I have no loyalty to either film or digital. Digital has it's headaches as does using and scanning MF trannies. <p>

 

I have done a couple more crops and again whilst not as pronounced as crops 1 and 2 (which must have been a bad scan) I feel that the digital file more than holds it's own. They are in my gallery. <p>

 

Has anybody else out there got any actual comparisons of their own to share? <p>

[neil l]

Steve must have been typing his response as I was mine. <p>

 

You make a lot of sense and I am starting to reallise that the Bronica/4990 combination was not the bargain I was looking for. If I were to make prints from the slide then it would no doubt be fantastic abut I am obviously losing a lot in the translation. <p>

 

If nothing else I feel I have a decision to make now!! <p>

[peter_nelson1]

<I>Neal Shields Photo.net Patron, may 25, 2005; 05:35 p.m.

As usual this is NOT "digital vs Film" it is digital vs scanner.

 

Film tests out to resolve 80 to 200 Lp/mm

 

Why do people persist in thinking that they can take a scanner that isn't capable of 1/5 that resolution and test film with it. </I><P>

 

I don't know what his scanner is, but a decent 4000 PPI film scanner such as my Nikon Coolscan V-ED is cheap and easy to use. I've never shot with any film that it couldn't easily resolve individual grain. (heck, my Canoscan 2710 could resolve down to the grain on plenty of common films!).

Furthermore, while it's true that <B>FILM</B> may be able to resolve to better than 100 lpm, very few <B>LENSES</B> can. The best 35mm primes can make it down to the 80-90 lp/mm range, medium format primes don't usually get better than 60-70 lp/mm. A top-quality zoom lens might also achieve that. And all of these are under ideal conditions of perfect focus, mirror lock, and cameras bolted to tripods.<P>

 

So consider that a modern DSLR such as the Canon EOS 20D packs 3504 sensor elements across 22.5 mm - that's 155 elements per linear mm, which means it can theoretically resolve over 70 lp/mm. This means that it's already theoretically better than most of the lenses it's used with.<P>

 

Of course in practice none of this applies due to AA filters on DSLRs, and camera and subject motion and poor focus in real world photography.

[larrydressler]

Why won't this forum allow attached or imbeded pictures?

[lee_kok_keong]

I use both digital and film, D70 and F2 I also own meddium and large format camera,

in most case I project my 35mm slide, and view on light table for med and large

format. I digitize my slide using Crossfield Magnascan 636. Those scanner you use is

just a toy. Film is much better in tone/colour and resolution. Digital is convenient and

not bad too. film is still easier to store and search, I have optical disc that no drive in

the market available any more, and have cd that fail to read. Is stupid to compare the

way you do. use a loup 25x or so and view your slide. In time to come digital may be

even better, however my 30 over years camera still serve me well, I dont think your

digital camera can last that long.

[oscar_van_der_velde]

I wonder which camera from N L has taken the photos he likes the best. A back-to-basic Bronica handles much differently compared to a full-featured digital SLR and this of course influences how you shoot. After 12 shots you have to change film again, so they'll better be good ones! The camera has to support your shooting style in order to enjoy taking photos with it.

 

By the way, I shoot also an old Bronica S2a and have a Canoscan 9950F. with strong sharpening applied, you will notice more detail and be very acceptable. It should be fine for 50x50cm prints and of course the web, which is what I want for now... and the tranparencies still hold their detail if I'd ever like to run them through a better scanner.