Ad
Ad
Ad
Pages: « 1 [2] 3 4 ... 7 »   Bottom of Page
Print
Author Topic: Interesting comparison of IQ180 and 8x10" film  (Read 30145 times)
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3550


« Reply #20 on: September 23, 2011, 02:57:12 AM »
ReplyReply

This is only the last in a series of tests of scans vs digital where the digital camera is the best that money can buy and the scan is made by a cheap junky scanner. In this case, it isn't so much the scanner itself, but the idea that you can't get any more than 745 ppi.

Hi Lenny,

I agree, to me this was the more surprising part of the test. Even despite the large format lens (of unknown quality), DOF, diffraction, and film flatness isues, I'd suspect that up to some 4000 PPI there would be some potential resolution that could be unleashed. A high quality film image that is not limited by diffraction will benefit from scans up to 8000 PPI (if only for grain structure), but whether the resolution was there in the images used, I don't know.

Maybe the authors did compare higher resolutions scans, or did an inspection with a strong loupe, but in that case they should have made that point more clear than just stating that scanning at a higher resolution wouldn't have helped. As the article is, there will be doubt as to the fairness of the comparison, which is a pitty. Maybe they can add some info about the lens and scan aspects, or at least use a higher quality scan, followed by proper downsampling.

Cheers,
Bart
Logged
qwz
Jr. Member
**
Offline Offline

Posts: 82



WWW
« Reply #21 on: September 23, 2011, 02:58:28 AM »
ReplyReply

Drum scanned (Dainippon screen SG 608) at 745 dpi it is joke?Huh

Acros and Ektachrome you must scan at least at 3000 dpi!

Logged
SCHWARZZEIT
Newbie
*
Offline Offline

Posts: 13


WWW
« Reply #22 on: September 23, 2011, 03:00:55 AM »
ReplyReply

The test as it is presented is not exploiting the full potential of the 8x10" format, not by a long shot. It's hard to say what went wrong on these shots but my experience with 8x10" has been quite different. What we're fed here is an embarrassment if it's supposed to be a point of reference of what can or cannot be done with an 8x10". Even a good 4x5" scan can look much better than this at 80 MP.

Some questions that came up:
Why did the author choose these selected crops? Why not offer the full resolution images as a download? Seeing the whole thing would give a much better comparison on the quality of the test images.
Why these odd display ratios? E.g. 200% displayed as 25%? Are we looking at 50% crops from the pixel level? Why not show plain crops on the pixel level?
Ok, on Markus' additional website the crops are presented at twice the size. It's quite obvious from these that the 8x10" shots were not properly focused on the cropped area. At this low magnification I'd expect an 8x10" to look much sharper if the shots were properly focused.
As Dan pointed out, why was the scan limited to 745 ppi which isn't even 80 MP? It's closer to 40 MP. If 745 ppi was the resolution used on the scanner and the pixel dimensions of the scan were 8874 x 7229 then the scanned area had to be larger than 8x10".
If I had seen that the films were not sharp through a 10x loupe on the light table I would have reshot the test or called it a failure. But publishing this as a reference for 8x10" is not really fair towards those readers who have never seen what a good 8x10" image looks like. It is indeed misleading.

I agree with Jack that it is very difficult to use the full potential of 8x10" and larger formats. But it's not that hard to get good enough quality to match only 80 MP if the camera is properly aligned.
A while ago I posted this sample showing some of the potential of what is possible with 8x10" color neg in terms of resolution.

As I've stated before I don't think the resolution of film or a film format can be described by a single number. It's much more complex than that. The resolution of film can be very high if you have lots of object contrast but it drops with lower object contrast. Of course the MTF of the taking lens matters to get as much of that contrast on film. But with a film format as large as 8x10" you should have plenty MTF if you only want to match 80 MP.

Diffraction at f/32 is not a problem on 8x10" at this magnification. With an Airy disc of 42 micron diameter you can resolve details as fine as 11 micron. In fact f/64 would be a closer match in DOF compared to the IQ180 at f/16. BTW, stopped down to f/16 the effect of diffraction can already be detected on the 5.2 micron sensor. If this was necessary to get the desired DOF then it would not have been possible to do any better on this shot with an 8x10" as the shot itself would have been ultimately DOF diffraction limited, and thus the 8x10" should have been used at f/64 or smaller. Sometimes you can get away with a larger aperture by using tilts.

Here is a full resolution Zoomify image of Geiranger Fjord (Norway) I shot in 2009 on 4x5" (lens: Schneider 110 XL at f/16.3; film: Fuji  Acros 100):
http://www.high-end-scans.de/img/bilder/zoomify/DV090809_01_6000ppi/DV090809_01_6000ppi.htm
Please note the full screen view button.

It's a 6000 ppi drum scan with a slight amount of smart sharpening for viewing on screen. After cropping the film rebate the image is about 650 MP. Yes, it's not tack sharp on 1:1 pixel level. Now the question is how well does an IQ180 or Aptus-II 12 file take an enlargement to 650 MP? Any non-stitched samples?

To keep in perspective of what you're looking at here's the same image reduced to about 80 MP and sharpened for the screen:
http://www.high-end-scans.de/img/bilder/zoomify/DV090809_01_res84/DV090809_01_res84.htm

-Dominique
Logged

qwz
Jr. Member
**
Offline Offline

Posts: 82



WWW
« Reply #23 on: September 23, 2011, 03:22:55 AM »
ReplyReply

Quote
The author clearly indicates higher resolution scanning would not help. Airy disk diameter at f/32 would be around 42 microns, lenses are said to perform at 20-30 lp/mm.
Scneider APO-Symmar 5.6 240mm
clearly has 60 lp/mm at least

Quote
f/11 76  76  60
f/16 67  67  54
f/22 60  60  54
http://www.hevanet.com/cperez/testing.html

You need quadriple ridicolous 754dpi on good scanner.
Logged
djoy
Jr. Member
**
Offline Offline

Posts: 50


« Reply #24 on: September 23, 2011, 04:14:17 AM »
ReplyReply

I was quite interested to see this article appear, but I'm afraid I can't take it seriously due to the poor way (IMHO) in which the scanning was handled.

745dpi? sorry, that's just not going to work.

It might have given a digital image of similar pixel dimensions as the IQ180, but it was not utilising all the information on the 8x10 transparency. I don't know anyone who would do a master scan like that.

I don't doubt for one moment the quality of the IQ180 is exceptional, especially if one goes to the trouble of using such a technically precise camera system as the Alpa and calibrates the sensor gap as has been demonstrated on this site. Similarly I don't doubt it is significantly more convenient that using 8x10 film and view cameras.

I also don't doubt that an 8x10 piece of film, well shot, contains a *huge* amount of data, much of which went unused in this comparison. Sad I've heard it said many times that the skill of the scanner operator is a significant factor in the quality of drum scans, I believe this test reinforces that. A better comparison IMHO would have been to scan the 8x10 at an optical resolution and aperture appropriate to the scanner and film grain, a good scanner operator would know the optimal limits for their scanner. At least that way all the data is used in both formats, a much fairer comparison. Then the scanned image could be down- sampled to the resolution of the IQ180 shot, and it might also be interesting to see the IQ180 shot up-sampled to the resolution of the scan.

I do shoot 5x4 (occasionally) and own a drum scanner, but I do not own either of the formats in this comparison, nor do I expect I ever will. Nevertheless I am still curious as to the difference in image quality between the two, but I'm afraid this comparison leaves me no wiser.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8874


« Reply #25 on: September 23, 2011, 04:19:35 AM »
ReplyReply

There's an interesting phenomenon at work in such tests, comparing smaller formats with larger formats, whether it's a comparison between the Canon G12 and a P45+, an 11mp Canon 1Ds with a 6x7cm film format, or an IQ180 with an 8x10" film format.

The question is, should all such comparisons be on an exactly equal playing field with regard to all the major parameters that we photographers are concerned about, such as FoV, DoF and (often overlooked) shutter speed.

Matching FoV is a given because that's so central to composition. Matching DoF is not something we're particularly fussy about, and matching shutter speed is just ridiculous.

I'll always remember that comparison between 35mm film and 4"x5" film made by Lars Kjellberg of Photodo fame.

He did his best to match DoF, comparing shots at F5.6 on 35mm with F22 on 4x5. He made a compromised attempt to get shutter speeds 'not hugely different' by using an ISO 100  B&W film with 35mm and an ISO 400 film with 4x5.

The result was, the 35mm shot was slightly sharper but still showed more grain, so one might have concluded, on balance, that the 4x5 shot was preferred.

But what would have been the result if Lars actually had equalised shutter speeds? He would have needed to push process the 4x5 shot to ISO 1000 or greater, and I suspect the 35mm shot would actually have been slightly better in all respects.

The lesson here is that different formats have different strengths under different conditions. If one attempts to get a completely equal playing field for the comparison, it is necessary to make the comparisons on at least two or three different playing fields in order to get an understanding of the comparative strengths and weakness of the different formats.
Logged
theguywitha645d
Sr. Member
****
Offline Offline

Posts: 970


« Reply #26 on: September 23, 2011, 09:05:02 AM »
ReplyReply

745dpi? sorry, that's just not going to work.

Why? Undersampling usually results in sharper looking images. The images are soft at 754dpi so the scanner is not undersampling. How is this image going to catch up to the digital image by simply adding pixels?
Logged
John Rodriguez
Jr. Member
**
Offline Offline

Posts: 91


« Reply #27 on: September 23, 2011, 11:07:12 AM »
ReplyReply


THIS.

I think all we're looking at here are DOF differences.  I'd like to see a similar test but at equal measured DOF for both systems OR using a target with a very shallow DOF window and optimizing each for diffraction (hell even a brick wall).   


Indeed!  Even if one has succeeded in getting perfect focus and perfect flatness of the film, one can't expect a lens to be tack-sharp at F32.

On the other hand, even F32 with 8x10" format does not provide the same DoF as F11 on the IQ180. The diagonal of 8x10" film is approximately 4.8x the diagonal of the IQ180 sensor, therefore for equal DoF one should use an F/stop number which is 4.8x greater, that is, somewhere between F45 and F64.

However, this tendency to use a shallower DoF with the larger format, when making comparisons with a smaller format, seems to be a fairly common practice, in order not to place the larger format at too much of a disadvantage.

It would be interesting to see comparisons also using f2.8 with the IQ180 and F11 with the 8x10 format. But maybe lack of film flatness would be too much of a problem.
Logged
lenny_eiger
Newbie
*
Offline Offline

Posts: 21


WWW
« Reply #28 on: September 23, 2011, 12:52:02 PM »
ReplyReply

As I've stated before I don't think the resolution of film or a film format can be described by a single number. It's much more complex than that. The resolution of film can be very high if you have lots of object contrast but it drops with lower object contrast. Of course the MTF of the taking lens matters to get as much of that contrast on film. But with a film format as large as 8x10" you should have plenty MTF if you only want to match 80 MP.-Dominique

This begins to hit the matter on the head. First of all, we all agree that 745 is a joke. We don't know whether it was undersampled or not as we don't know the aperture of the scanner, which is what controls the under/over sampling vs the spi. I will state categorically that the IQ80 can not stand up to what an 8x10 can do, if that's the question people are trying to answer. The test is a fools errand.

Consider this, you have a 4x5 image with a telephone pole in it. The pole is one of those with a thousand shades of reddish brown. It is 1/4 of an inch wide on the film. Compare the 1/4 square of a piece of this pole to an 8x10 of the exact same shot. You now have 4 times the amount of film with which to describe all of these tonal differences. There will be, there is no question, much more tonal information. At 6x7 size, our .25 inch square pole snippet becomes a .1375 inch square, about half the size.

Now if one doesn't care about the tones in the image, you're shooting lith film or something, or making really contrasty prints, then this discussion of resolution becomes reasonable But most of us aren't doing that.

I have been working on understanding all the edges of this scanning stuff for many years. What I can say is that it is much better than the numbers will allow. I am usually scanning 810's at 2666 spi, a number that equates to 568 megapixels (a far cry from 80). I know there isn't enough resolution to suggest that when and if they ever make a 568 mpxl camera, that it will equate. However, it is much further than the 80. And it is patently ridiculous, when my scanner is rated at an optical resolution above 7300, to suggest that one can only get 745. The stuff coming out of my scanner is razor sharp... at these higher resolutions, and I have made the 20 foot prints to prove it.

Lenny
EigerStudios
eigerstudios.com
Logged

EigerStudios
Museum Quality Drum Scanning and Printing
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7308


WWW
« Reply #29 on: September 23, 2011, 02:16:10 PM »
ReplyReply

Hi,

OK, you have images to prove your point, why don't you post them?!

Actually, I'm somewhat skeptical.

745PPI seems like to low res to me, on the other hand the 745PPI scan clearly shows a lot of softness. Increasing resolution would just cause more softness.  My guess may be that stopping down to f/32 seriously limits resolution. According to my testing on DSLRs going from f/8 to f/22 cuts linear resolution in half, turning a 24MP camera into a 6 MP camera. So my guess is that an 8x10" camera at f/32 would be able to produce 1/4 to 1/6 of the resolution achieveable at f/11. To this comes film flatness and film to ground glass alignment.

One of the niceties with digital is that you have immediate feedback, if you botch it you can check directly, with film you see it when it's back from the lab.

Best regards
Erik


This begins to hit the matter on the head. First of all, we all agree that 745 is a joke. We don't know whether it was undersampled or not as we don't know the aperture of the scanner, which is what controls the under/over sampling vs the spi. I will state categorically that the IQ80 can not stand up to what an 8x10 can do, if that's the question people are trying to answer. The test is a fools errand.

Consider this, you have a 4x5 image with a telephone pole in it. The pole is one of those with a thousand shades of reddish brown. It is 1/4 of an inch wide on the film. Compare the 1/4 square of a piece of this pole to an 8x10 of the exact same shot. You now have 4 times the amount of film with which to describe all of these tonal differences. There will be, there is no question, much more tonal information. At 6x7 size, our .25 inch square pole snippet becomes a .1375 inch square, about half the size.

Now if one doesn't care about the tones in the image, you're shooting lith film or something, or making really contrasty prints, then this discussion of resolution becomes reasonable But most of us aren't doing that.

I have been working on understanding all the edges of this scanning stuff for many years. What I can say is that it is much better than the numbers will allow. I am usually scanning 810's at 2666 spi, a number that equates to 568 megapixels (a far cry from 80). I know there isn't enough resolution to suggest that when and if they ever make a 568 mpxl camera, that it will equate. However, it is much further than the 80. And it is patently ridiculous, when my scanner is rated at an optical resolution above 7300, to suggest that one can only get 745. The stuff coming out of my scanner is razor sharp... at these higher resolutions, and I have made the 20 foot prints to prove it.

Lenny
EigerStudios
eigerstudios.com

Logged

Faintandfuzzy
Newbie
*
Offline Offline

Posts: 21


« Reply #30 on: September 23, 2011, 02:49:45 PM »
ReplyReply

Erik, considering Lenny's experience in this area, I'll take his word for it.  As to this "test"....I'd say it ranks right up there with the D30 vs Provia and the 1Ds vs 6x7 film.....two tests which have been debunked and laughed at for the better part of a decade.

The part I don't get is the 1Ds vs 6x7 film.  A couple of years after the comparison was 645 vs the 1Ds2.  The confusing part was how it was claimed 11mp beat 6x7 film....yet 17mp "almost" matched 645.  Anyone see a glaring problem there.

No, this test, like the others, were made using poor methods and questionable outcomes.  In other words, a waste of time.
Logged
AFairley
Sr. Member
****
Offline Offline

Posts: 1137



« Reply #31 on: September 23, 2011, 02:53:03 PM »
ReplyReply

If one is interested in the absolute level of information capture between the two media, a quick and dirty way to take the scanner issue out of the equation (without rescans) would be to shoot macros of the films with a good DSLR, no?  8000 dpi is a piece of cake that way.
Logged

SCHWARZZEIT
Newbie
*
Offline Offline

Posts: 13


WWW
« Reply #32 on: September 23, 2011, 03:04:46 PM »
ReplyReply

Somehow I forgot to include the link to the thread where I posted the crops of my 8x10" sample.

-Dominique
Logged

feppe
Sr. Member
****
Offline Offline

Posts: 2909

Oh this shows up in here!


WWW
« Reply #33 on: September 23, 2011, 03:13:09 PM »
ReplyReply

OK, you have images to prove your point, why don't you post them?!

Actually, I'm somewhat skeptical.

Erik, you might want to google Mr Eiger and his credentials before being skeptical of anything he says when it comes to photographic image quality, especially scanning.
Logged

Faintandfuzzy
Newbie
*
Offline Offline

Posts: 21


« Reply #34 on: September 23, 2011, 03:13:39 PM »
ReplyReply

Somehow I forgot to include the link to the thread where I posted the crops of my 8x10" sample.

-Dominique

Looking at your crops, the 1440 ppi have a lotr of detail.  But you can still get more which is obvious in the 2400ppi samples....where the blinds pop in as do the flowers.  There is no way one could claim that 745ppi will get all the detail when we can clearly see benefits out to 2400ppi at least.

Thanks for posting this.  I further debunks the "test"
Logged
Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2823



WWW
« Reply #35 on: September 23, 2011, 04:03:06 PM »
ReplyReply

Why? Undersampling usually results in sharper looking images. The images are soft at 754dpi so the scanner is not undersampling. How is this image going to catch up to the digital image by simply adding pixels?

seems like over sampling and then downrezzing to a similar dimension offers sharper images.  Isn't this a pretty typical workflow for 8x10 film ... scan at very high resolutions and then resize downward for output?
Logged

lenny_eiger
Newbie
*
Offline Offline

Posts: 21


WWW
« Reply #36 on: September 23, 2011, 04:40:07 PM »
ReplyReply

745PPI seems like to low res to me, on the other hand the 745PPI scan clearly shows a lot of softness. Increasing resolution would just cause more softness.  My guess may be that stopping down to f/32 seriously limits resolution. According to my testing on DSLRs going from f/8 to f/22 cuts linear resolution in half, turning a 24MP camera into a 6 MP camera. So my guess is that an 8x10" camera at f/32 would be able to produce 1/4 to 1/6 of the resolution achieveable at f/11. To this comes film flatness and film to ground glass alignment.

This is incorrect. It's a low end scanner. Softness, given good sharp film, is a function of scanning aperture and quality of scanner. Increasing the spi would not increase softness. It is simply ridiculous to imagine that 745 would match 4000.

Further, digital lenses cut their resolution much lower down the scale than large format lenses. I just tested my 150 Sironar S and the thing was extremely crisp at f45. Sorry, f22 wasn't any better.

One of the niceties with digital is that you have immediate feedback, if you botch it you can check directly, with film you see it when it's back from the lab.

That's true. That's the only benefit of digital however. Those of us that have been doing photography for long enough have worked on our processes long enough to get it right. I almost never have an issue with exposure and development.

Lenny

EigerStudios
Logged

EigerStudios
Museum Quality Drum Scanning and Printing
dreed
Sr. Member
****
Offline Offline

Posts: 1221


« Reply #37 on: September 23, 2011, 05:46:00 PM »
ReplyReply

That's true. That's the only benefit of digital however. Those of us that have been doing photography for long enough have worked on our processes long enough to get it right. I almost never have an issue with exposure and development.

And having visited one of Ken Duncan's galleries and stared in awe at the quality of prints (very large prints) that he makes from film, I can't disagree with what you've said about those that still work with film. There are other large format prints that I've seen in galleries where a 6'x6' (or similar sized print) is hanging that was made from an 8x10 and you're left wondering how do they do it...
Logged
lenny_eiger
Newbie
*
Offline Offline

Posts: 21


WWW
« Reply #38 on: September 23, 2011, 06:01:00 PM »
ReplyReply

And having visited one of Ken Duncan's galleries and stared in awe at the quality of prints (very large prints) that he makes from film, I can't disagree with what you've said about those that still work with film. There are other large format prints that I've seen in galleries where a 6'x6' (or similar sized print) is hanging that was made from an 8x10 and you're left wondering how do they do it...

Truly, its a lot of work. Making a great print with inkjet is the same it takes to do it in the darkroom. Everyone thinks its automatic because its digital, but it isn't. Basically I print something, then I look at it. If I can improve it, I make another print. And so on.... Knowing when you can improve something takes knowing yourself, as well as knowing what's possible. There's no substitute for studying one's Photohistory, working with others' whose opinion you value and taking the requisite years to figure it all out.

Lenny

EigerStudios
Logged

EigerStudios
Museum Quality Drum Scanning and Printing
Ray
Sr. Member
****
Offline Offline

Posts: 8874


« Reply #39 on: September 23, 2011, 06:56:38 PM »
ReplyReply


Further, digital lenses cut their resolution much lower down the scale than large format lenses. I just tested my 150 Sironar S and the thing was extremely crisp at f45. Sorry, f22 wasn't any better.


Hi Lenny,
That interesting! Is this the lens?  http://www.bhphotovideo.com/c/product/43853-USA/Rodenstock_160702_150mm_f_5_6_Apo_Sironar_S_Lens.html

It does mention in the overview that the ideal Working Aperture Range is f/11 -22.

What's this lens like at F11?

My understanding is a manufacturer will optimise a lens to produce best results over a particular range of apertures which are likely to be the most useful in relation to the format of camera the lens is designed for.

As far as I know, we have not yet defeated the laws of diffraction, although we are apparently working on it, constructing artificial materials through processes of nanotechnology.

The essential problem with all such comparisons involving different formats of cameras, using different lenses which may also have different performance characteristics even at the same aperture, is reconciling the test methodology with the principle of "best tool for the job".

It is a given, in the real world, that any good photographer will not only choose the particular format of camera that best suits his purposes, but also use his chosen format, and the selected lens that goes with it, in a way which produces the best results.

In my view, the purpose of these comparisons between different formats and types of cameras, is to learn in what circumstances one format of camera may produce better results than the other.

Now it seems very clear to me, if the 8x10 format used at F32 cannot even closely match the resolution of the IQ180 used at F16, there's not much point in equalising DoF by using the 8x10 at F64. The differences would be even wider.

I think the results of this test are very clear, that in situations where a reasonably long DoF is desired, the IQ180 without a shadow of a doubt produces better image quality than 8x10 film (at least the film types in the test).

What is missing from this test is a comparison at the shallow end of DoF. For example the IQ180 at F2.8 versus the 8x10 format at F11 or F13. Would this situation favour the 8x10? Somehow, I doubt it, but it would be good to see a comparison, perhaps of a still-life taken in the studio.

I notice that Markus has rescanned the results at a higher resolution which has had the effect of narrowing the differences slightly, but not changing the over all conclusion.

Regards

Logged
Pages: « 1 [2] 3 4 ... 7 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad