80mp digital better than 8x10?

[paulr]

Calculating theoretical megapixels from scans rellies on a mountain of fallacies. It just doesn't work that way. My black and white 4x5 scans are 56 megapixels (after down-sampling to my printing resolution). And yes, the results look great at the sizes I print. But I know from my recent digital experience that comparable results could be had from fewer than half that number of pixels, if they came from a good sensor.

A lot of this has to do with the MTF characteristics of film, a lot has to do with s/n performance of the film + scanner vs. the digital sensor. I don't know what all the factors are. But I do know that a high quality 80mp sensor would be capable of stupefyingly detailed and sharp results.

[John NYC]

Calculating theoretical megapixels from scans rellies on a mountain of fallacies. It just doesn't work that way. My black and white 4x5 scans are 56 megapixels (after down-sampling to my printing resolution).

Sounds about right. That would give you a 27.9 x 22.3 inch print at 300dpi from your scan.

[tgtaylor]

Doug Peterson, nice to see you join the discussion.

Can you point to any more credible and realistic comparisons between the resolving power of the IQ180 and large-format film than the recent most dreadful one on Luminous Landscape which started this discussion?

And do I have it right from the Wikipededia pixel specs for the sensor (10328x7760) and my back-of-the-envelope calculation (using the sensor size of 53.7mm x 40.4mm) that at the base ISO--without considering binning-- the sensor can theoretically resolve 96 lp/mm?

Doug hasn't replied yet but the theoretical resolution of 96 lp/mm for the IQ180 sounds correct to me. If that is correct then that puts the published resolution of film in the same ball park as the IQ180. For example Fuji's data sheet states the resolution of its Velvia 100F to be 80 to 160 lines/mm depending on the contrast of the chart. Running somewhat concurrently with this thread, a thread on the apug site is discussing the resolution of film and several posters there are claiming that the published resolution of a number of films can be significantly higher according to the developer used.

Thomas

Thomas

[John NYC]

For example Fuji's data sheet states the resolution of its Velvia 100F to be 80 to 160 lines/mm depending on the contrast of the chart.

Find me an 8x10 lens that resolves anywhere near that across its entire image circle.

[tgtaylor]

Find me an 8x10 lens that resolves anywhere near that across its entire image circle.

Good point John. In the 3d edition of the Darkroom Cookbook, Steve Anchil states that all modern lens are able to outresolve film. So I did a quick search on the B&H website for the resolution specification of the new digital lens and found only phrases like "Extremely-High Resolution/Contrast" but no specs. Likewise I searched the Schneider website for the resolution specs of my newest lens - a 360mm Schneider Symmar-S and also came up with a blank.

Thomas

[TJV]

Well, from where I sit today an IQ180 sounds pretty good – just got a call from the lab saying they accidentally destroyed 30 sheets of my film in a processing accident. Can't even measure the resolution on them, just hope they somehow become iconic like R Capa's D-day photos...

[thrice]

Here is a crop from a 2400 dpi scan resampled to 745 dpi this is 4x5 and Kodak E100G, using my Nikkor-SW 75/4.5.







Even resampled to 8774pixels on the long end (same res as their 8x10 scan) it looks better than theirs.

[thrice]

Well, from where I sit today an IQ180 sounds pretty good – just got a call from the lab saying they accidentally destroyed 30 sheets of my film in a processing accident. Can't even measure the resolution on them, just hope they somehow become iconic like R Capa's D-day photos...

I can imagine nothing worse... *bro-hug*

[rdenney]

I've seen Kirk mention several times here on the forum that the transition from film to digital didn't hurt his profitability. Maybe he can chime in here and explain.

I'd like to hear that, too. But I suspect that Kirk's and Brian's business models have been rather different. Kirk is an architectural photographer primarily, and thus mostly does field work. Brian's work in New York was of a different nature, and with a different set of clients and their expectations.

For the work that I did, film was included as part of the price, but in contracting terms, I did firm fixed price contracts: A given product for a given quantity of dollars. The reason I did that work is because my clients didn't trust my efficiency, and thus used that contracting method to put the risk of being inefficient on me.

For event work, the model in the old days was to charge a firm fixed price for showing up and then providing a proof book, followed by charging for enlargements by the print. Boy, did we mark up those prints! That model seems to be gone now.

Brian's situation is quite different. As a known quantity in the New York scene, he could have perhaps built his contracting model around a time-and-materials method, where he charges by the hour, plus expenses. And he could have charged a markup on expenses. In the contracting world, we'd call that "general and administrative" expenses, marked up as a fixed percentage of expenses (including sub-contractors). You have to be at the top of the contracting world to be able to demand those terms, of course.

They may also be the terms demanded by his clients, versus the terms acceptable to Kirk's clients. The market sets the price...

Rick "speculating on Brian's business model from the perspective of quite a bit of experience with contracting" Denney

[rdenney]

A lot of this has to do with the MTF characteristics of film, a lot has to do with s/n performance of the film + scanner vs. the digital sensor. I don't know what all the factors are. But I do know that a high quality 80mp sensor would be capable of stupefyingly detailed and sharp results.

Yes. When we report seeing additional detail up to about 2400 spi on, say, the Epson flatbed scanner, we are reporting perhaps 10% MTF (minimum discernable signal) for subject material that the lens might have delivered at 50% MTF or more. The sensor might be much closer to what the lens delivers. But my experience with digital suggests that there is a floor beneath which the digital sensor drops precipitously to zero. Film provides a more gentle fall. That's why digital loses it more profoundly when over-enlarged compared to film. And that's why tools to allow over-enlargement of digital files do so by introducing random noise (e.g. Genuine Fractals)--it rounds the corner of that cliff edge.

But if we don't over-enlarge, then both methods produce quite good MTF. Different scanning methods impose different definitions of "over-enlarge". I have made prints up to more than 10x from my Nikon scannner with good results, but I don't think I'd be happy with prints larger than maybe 6x from the Epson, and that's probably more than what many would accept. But I routinely make very good looking prints from digital cameras at 20x.

8x10, of course, has a bit better than a factor of 4 advantage over the digital back from the start because it's about that much bigger. But at least half of that is given up using the scanning method most of us have available to us economically, for routine work.

It seems to me that if we are constrained by digital technology and sensor cost to large enlargement ratios, that puts a greater burden on everything else in the chain. Thus, we end up demanding far more expensive lenses of newer design. One key advantage to me for the larger format is that I can use old lenses effectively for the print sizes I use. That's a key cost issue. Consider the price of the Hasselblad kit as Brian outlined above. That's far more than most of us could possibly consider (which was Brian's point). And we need the lens diversity to make use of the full format of the sensor--giving up big chunks of the sensor through cropping gives up big chunks of enlargeability, given the sharp cutoff.

Rick "observing that delivering 100 lpm costs much more then twice what delivering 50 lpm costs" Denney