Conspiricy Theory

[Prof_Pixel]

The Dianippon drum scan resulted in a 64 megapixel image vs the 80 megapixel digital image (and is a good scanner); that's not my problem with the test. The aspect ratios are different so the real difference in size is less - the film image would be 70 megapixels.

Like I said " All you can really say about the results is that they apply to the two systems tested". There were lots of variables that weren't controlled.

 

[E. von Hoegh]

The Dianippon drum scan resulted in a 64 megapixel image vs the 80 megapixel digital image (and is a good scanner); that's not my problem with the test. The aspect ratios are different so the real difference in size is less - the film image would be 70 megapixels.

Like I said " All you can really say about the results is that they apply to the two systems tested". There were lots of variables that weren't controlled.

For starters, look at the inadequate tripod they put the 8x10 on. From there, it just gets worse. (Not to mention the scans...)

 

[Prof_Pixel]

For starters, look at the inadequate tripod they put the 8x10 on. From there, it just gets worse. (Not to mention the scans...)

The tripod is one of the uncontrolled variables - but don't get too hung up on the 80 megapixel vs 70 megapixel issue.

 

[E. von Hoegh]

The tripod is one of the uncontrolled variables - but don't get too hung up on the 80 megapixel vs 70 megapixel issue.

Oh, I'm not. 70mp is a fair equivalent of a 2 1/4 x 2 1/4 negative, not an 8x10 negative. The whole thing was just pointless from the point of view of a valid comparison. Just an example of a severely biased review, in this case an advertising puff for the digital back.

 

[removed account4]

for a long while there was one magazine published here in the states
which caters to people who use large format cameras &c
and every month or 2 months or whenever the magazine was published
there were countless complaints of poor image reproduction.
i think it was spread equally over both film capture and electronic capture ...

i don't think there is a conspiracy, just poor image reproduction and tight budgets.

 

[Prof_Pixel]

My concern is not with the scan - I suspect the 8x10 was not as sharp as it could have been - film plane flatness and tripod are just two issues; there are lots more.

BTW, I don't see anybody talking about how GOOD the digital files were ;-) just how bad the film files were.

 

[Prof_Pixel]

i don't think there is a conspiracy, just poor image reproduction and tight budgets.

I agree 100%

 

[MDR]

The IQ is a superb digiback no doubt about but it does not beat a well made 8x10. At f32 the depth of field would counteract the filmflatness problem apparently it didn't. The scans are something I would expect from a flatbedscanner and not from a drumscan. The IQ shot was sharpened in capture one (a superb programm maybe the best raw converter on the market imho), the scans of the film were sharpened with a homespun algorithm that supposedly reduces sharpening artefacts come on give me a break. The guy who's done the test is a 100% digiguy in and owns a company that specialises in fine art digiprints, so one can'T talk about him not beeing biased. A 100% film guy would be biased as well, so a photographer who is at home in both worlds would have been a better choice and is probably more objective.

Dominik

 

[Worker 11811]

A 100% film guy would be biased as well, so a photographer who is at home in both worlds would have been a better choice and is probably more objective.

Don't you think the best test would be made by a film guy taking the 8x10 film image and a digital guy taking the digital image then handing both images over to a third party for analysis?

Not only would that eliminate bias from the photographers, it would allow the reviewer to concentrate on the task at hand, reviewing, instead of taking pictures.

(In a perfect world... :confused: )

 

[MDR]

Don't you think the best test would be made by a film guy taking the 8x10 film image and a digital guy taking the digital image then handing both images over to a third party for analysis?

Not only would that eliminate bias from the photographers, it would allow the reviewer to concentrate on the task at hand, reviewing, instead of taking pictures.

(In a perfect world... :confused: )

You might have a point.

 

[Prof_Pixel]

Unless you could put the film and sensor in exactly the same camera, you can never have a completely 'fair' test. Like I've been saying, the best you can do is compare two systems meaning someone can always find fault with the test.

Shoot what you like and forget about trying to 'one-up' the other guy.

 

[SkipA]

Conducting poorly designed or purposely flawed tests is pointless unless your intent is use the results to mislead people. I think Worker 11811 is right. To eliminate as much bias as possible, a well designed test would try to eliminate as many variables as possible. Having multiple independent third parties compare the best efforts of the same scene photographed at the same time by expert analogue and digital photographers would go a long way toward making the exercise fair and the results more meaningful.

 

[Prof_Pixel]

Like I said - you can only compare systems - you can't generalize the results.

BTW, In my opinion, I would have conducted the test using a 4x5 system rather than an 8x10; lens quality would have been more comparable, film plane flatness would be better, less camera shake, etc.....

 

[lxdude]

The Dianippon drum scan resulted in a 64 megapixel image vs the 80 megapixel digital image (and is a good scanner); that's not my problem with the test.

I don't see why that would not be one of your problems with the test. 745 dpi? Visible jaggies?

745 dpi (I prefer spi, but whatever) cannot image anything less than 35 microns, fer cryin' out loud. A 745 dpi scan of a 24X36mm image yields less than .833 MP.
The writer claims the appearance of grain at that level means no more resolution is available. Has he heard that scanning technique can affect apparent grain in the scan? Has he heard of grain aliasing? Why does that Acros grain look so weird?


745dpi = 372.5 lp/inch. That converts to 14.66 lp/mm. Ridiculous.


The test purports to compare resolution. The scanning sampling rate negates any possible objective comparison, even a casual one.

The other issues regarding this test which you raised are also valid, and reveal this "test" to be fatuous.

 

[Prof_Pixel]

I don't see why that would not be one of your problems with the test. 745 dpi? Visible jaggies?

745 dpi (I prefer spi, but whatever) cannot image anything less than 35 microns, fer cryin' out loud. A 745 dpi scan of a 24X36mm image yields less than .833 MP.
The writer claims the appearance of grain at that level means no more resolution is available. Has he heard that scanning technique can affect apparent grain in the scan? Has he heard of grain aliasing? Why does that Acros grain look so weird?


745dpi = 372.5 lp/inch. That converts to 14.66 lp/mm. Ridiculous.


The test purports to compare resolution. The scanning sampling rate negates any possible objective comparison, even a casual one.

Don't get fixated on the scanning resolution; what is important is the total number of pixels captured for the scene. A 4x5 would have had to be scanned at 1500 ppi to generate the same number of total pixels while a 35 mm frame would have had to be scanned at something like 6000 ppi!

In the test, any give area of the scene (like a license plate) is reproduced by approximately the same number of pixels with both systems. The fact is the 8x10 image wasn't sharp (for reasons already mentioned and probably others).

 

[E. von Hoegh]

Hey guys, you can read all about it over at LFPF. One thread ran 32 pages.....

 

[Felinik]

Working with marketing in a similar industry selling gadgets to professionals and wanna-bee professionals, this is how it works; You pay for some kind of advert plan (x months reoccurring) included in the price you get some kind of editorial material about your product/company published, sometimes you just get a slot for an editorial and you produce it yourself (of course in collaboration with the publisher, but the less work for them, the better they like it...).

Go figure.

 

[dslater]

IMO scanning the 8x10 film makes this test pointless period. Once you do that, you're just comparing 2 different digital workflows. Assuming your final product is a print and not some kind of digital media, the only comparison I'm interested in is the final print. Make a couple of 30x40 or larger prints - use a fully wet process for the 8x10 film and a digital process for he digital image - ideally using an RA-4 paper based printer to avoid differences between Inkjet vs. RA-4.

After all who cares which one looks better when you're pixel peeping on some computer screen. What matters is which one produces a better print.

 

[Klainmeister]

I have to agree. I am concerned with a print, not a file.

 

[Mainecoonmaniac]

I would consider the source. I'm sure the writer is enamored with anything new. I would have to say there are people here that are anti-digital on this forum too. I would come to my own conclusion about the truth between digital and analog. Pardon the pun, but I'd be very wary of people with black and white views. The truth exist in shades of gray.

 

[Prof_Pixel]

What I'd like to see is a film comparison between a 'well shot' 4x5 and a 'well shot' 8x10 at the same final print size. I suspect the 4x5 would be sharper because of issues like lens quality and film plane sharpness.

 

[pbromaghin]

In the test, any give area of the scene (like a license plate) is reproduced by approximately the same number of pixels with both systems.

But this is precisely what I see as the most glaring error. In order to squeeze the license plate from the 8x10 down into the same number of pixels, they had to throw out (SWAG warning: Scientific Wild-Ass Guess coming) 90% of the data that represented the license plate. When the then they turn around and enlarge the license plates onto equally-sized printed areas, they compare a 10% sampling of the 8x10 data with 100% of everything else.

 

[Prof_Pixel]

> They had to throw out (SWAG warning: Scientific Wild-Ass Guess coming) 90% of the data that represented the license plate.

Sorry, but it doesn't work that way.

 

[pbromaghin]

Prof_pixel, I'm open to being educated. I would love to know what's wrong with my image of the process. But that probably goes too far into hybrid workflow for this forumn. Is there a good single source I can use to do so?

 

[Ian Grant]

What I'd like to see is a film comparison between a 'well shot' 4x5 and a 'well shot' 8x10 at the same final print size. I suspect the 4x5 would be sharper because of issues like lens quality and film plane sharpness.

I have shots showing the two formats quality and the difference is quite distinct, the 10x8 shots are way ahead and that's despite the 5x4 images being shot with a modern 1980's MC 150mm Rodenstock Sironar N lens and the 10x8 with a pre-WWII 12" Dagor (factory coated after WWII). Same tripod.

Every negative shot with my 10x8 cameras is higher quality than an equivalent made with one of my 5x4's. My normal print size is the same for both 5x4 & 10x8.

Ian