Sony AR7IV vs GFX 50 for DSLR scanning.

[andylynn]

No, it's also about a capture technology called "pixel shifting" as asked, a subject you haven't tried and call mumbo jumbo and a "trick".

Please read the context. The question was about using the tech for scanning. It’s in the title. You look like you’re encouraging a guy to get a new camera to improve on already overkill 50MP film scans. That’s terrible advice that you’re pushing with some tech jargon that is not about anything that is helpful in the context of whether a new camera will offer an improvement over 50MP scans.

 

Let me hazard a guess. You own an A7R mark whatever and you think everybody should be on team Sony.

 

I understand you and disagree with you.

[digitaldog]

Please read the context. The question was about using the tech for scanning. It’s in the title. You look like you’re encouraging a guy to get a new camera to improve on already overkill 50MP film scans. That’s terrible advice that you’re pushing with some tech jargon that is not about anything that is helpful in the context of whether a new camera will offer an improvement over 50MP scans.

 

Let me hazard a guess. You own an A7R mark whatever and you think everybody should be on team Sony.

 

I understand you and disagree with you.

Stop guessing; your assumptions, misunderstandings of technology and lack of experience is quite clear.

 

Yes babuophoto I know Jim and agree he'd be a good person to ping and I'll do so. Jim is experienced and knowledgeable about technology and does his due diligence in testing.

[babouphoto]

Thanks

 

Stop guessing; your assumptions, misunderstandings of technology and lack of experience is quite clear.

 

Yes babuophoto I know Jim and agree he'd be a good person to ping and I'll do so. Jim is experienced and knowledgeable about technology and does his due diligence in testing.

[andylynn]

Stop guessing; your assumptions, misunderstandings of technology and lack of experience is quite clear.

 

Yes babuophoto I know Jim and agree he'd be a good person to ping and I'll do so. Jim is experienced and knowledgeable about technology and does his due diligence in testing.

You’re assuming I don’t know anything about pixel shift. I’ve used it many times. Olympus had it when I was still shooting M4/3. I know it is a way to increase the amount of image data you collect.

 

It is completely unnecessary and a waste of money and time for a person who has a 50MP camera to buy a new camera in order to use pixel shift when scanning film, unless you’re shooting medium format 25 ISO B&W film or something.

 

Pixels are a wonderful thing to fanboy over but what you’re talking about does not provide a benefit for the thing that is the subject of this thread. If stringing together big words and pretending that anybody who disagrees with you is an idiot makes you feel better about yourself, go ahead, it doesn’t matter, but don’t be surprised when people correct your bad advice.

[digitaldog]

First, when you call a technology a 'trick' and mumbo jumbo, then you assume people are recommending a purchase of a camera (none have) but rather actual testing which of course you haven't done nor provided, you're not to be taken seriously. When you come here and state you've got a 'pro trick' without a lick of evidence you're a pro at anything, let alone photography, you're not to be taken seriously. When you call people fanboy's when they simply make factual statements about technology, you're not to be taken seriously. When you post as an anonymous troll, you're not to be taken seriously.

Here's the advise I gave to the OP you call bad: Do some testing and don't listen to trolls who haven't but form opinions that shouldn’t be taken seriously. It's the later part of this advise that apparently ruffles your delicate feathers and provides a need for you to continue assuming and maligning technology you don't understand and why again, you're not to be taken seriously. Understand?

I didn't think so.....

"Some people will never learn anything well, because they understand everything too soon." -Source Unknown

[andylynn]

Ok, I understand that it makes you feel better and I’ve said what I wanted to about how insane it would be to get GAS over whether one can scan at more than 50MP. I’m going to let the rest of it go because you’re not important enough to argue with and I have to go do some errands.

[digitaldog]

Thanks for clarifying again your misunderstandings about resolution and color image data. On your errands, consider the not insane concept of making an image....:)

[rodeo_joe1]

Hmmm. Simply referring to another person's blog, while berating other people for providing no 1st hand evidence and not having any personal experience. Does anyone else notice the irony here?

 

So, how to show the true grain structure of a piece of film in a digital format? Well, I have some very sharp enlarged prints 10x12" that I also have the negatives to.

How about a flatbed scan of such a print, compared to a direct digital camera copy from the same negative.

 

This particular example is a 6x6cm FP4 negative, slightly cropped:

I figured a B&W negative would more easily show any colour mottling due to Bayer aliasing.

 

Here's the comparison:

The flatbed scanner was a good quality tri-line CCD model, capable of much higher resolution than the 600 ppi used here. It would render a 7200 pixel square (~ 50 megapixel) image from the whole negative at this scale. The camera used was a relatively cheap Sony 24 megapixel a6000, set up to capture the same area of the negative at close to the same RR. No resizing was done to either sample. They're 'straight' colour shots from the camera/scanner with only some sharpening applied to the flatbed scan.

 

If I look critically, I can see some slight colour mottling in the camera copy. Whether this is due to Bayer interpolation, or just the fact that the camera copying setup renders the negative grain more sharply, I really can't say. To me, it's not a major issue, since such close examination of the image would require ones nose being pressed close to a 5 foot square print.

 

For completeness, here's the grain structure of the film up close and personal.

[digitaldog]

The first hand example NO one has provided is a capture of a slide with and without Pixel Shifting on an appropriate sample (fine ruled lines or fabric would be ideal). Two so called "blogs" did show examples of captures side by side with and without Pixel Shifting and the differences are visible to me! But this wasn't a test of slides so my only take away is that in some examples, this technique absolutely isn't a trick let alone mumbo jumbo. So no, I don't see any irony in those specific examples of this capture technology on one kind of capture.

 

How will the OP actually know if this technique improve capture of slides? He can as he suggested, rent a camera and conduct his own tests (a move I applaud and recommend), find actual examples from those that have (none yet but perhaps a possibility), or listen to those who make recommendations without any sort of testing (a move I don't believe has any merit).

 

It is entirely possible that there will be no benefit of the camera or optional capture process the OP asked about. But thus far, there is no proof it will not so a well structured test is in order for those who wish to base opinion on FACTS.

 

I fully accept this takes work and it is easier to provide technical "advice" from speculation!

[babouphoto]

I'll just weigh in. No one has done macro tests and none on B&W film (not that it should matter in theory) and none for scanning specifically. I'll just rent the AR7 or just keep going with the GFX which is working great so far. Either way,I guess we can stop the rants unless someone has done a test ;-) The closest so far has been Jim Klasson's extensive testing. Thanks everyone!

[rodeo_joe1]

and none on B&W film (not that it should matter in theory)

Ah, but B&W film is exactly where Bayer filter 'aliasing' will show up most. Same as with B&W Siemens star patterns and other high contrast high spatial frequencies.

 

That's the very reason I chose a B&W neg for the comparison above.

 

The fact that there should be no colour at all in the image makes any colour fringing, mottling or moire all the more obvious. Whereas a colour image tends to mask the effect, especially dye cloud patterns in film. The random scatter of small coloured 'spots' is an ideal place for other coloured spots to hide.

 

However, since the effect takes place at pixel level, and pixels are very tiny these days, it often doesn't make itself obvious.

 

There are subtle errors and less-than-perfect aberrations in most technologies, which once you know they exist can be irksome. "Ignorance is bliss" is a good truism, and there's no point looking too hard for 'faults' that can't be noticed unless you look too hard.

[digitaldog]

Gents - if someone knows Jim Klasson maybe they can ask him to weigh in. I love his blog and posts. He is very technical and provides detailed analysis. His review of Pixel Shifting is what got me started with this topic in the first place!

 

Here's what Jim emailed me:

 

I've not done any slide copying with pixel shift, but I think it could work quite well if the setup is rigid, there is no vibration, and the slide isn't warming up enough to change the flex during the exposure sequences.

 

Jim

So yeah, do your own testing ideally from a rental, then you'll know....

[digitaldog]

And he (Jim) just rewrote back this:

The more I think about it, the better pixel shift sounds for slide scanning. It could virtually eliminate the chromatic element of "grain aliasing". It would be interesting to see some a/b comparisons. If I'm right about grain aliasing, it would be most effective with higher ISO slide film. I don't think I'll try this myself, since I have no need to scan old slides right now, and I haven't made any new ones in at least two decades.

So yeah, do your own testing ideally from a rental, then you'll know....

[babouphoto]

And he (Jim) just rewrote back this:

The more I think about it, the better pixel shift sounds for slide scanning. It could virtually eliminate the chromatic element of "grain aliasing". It would be interesting to see some a/b comparisons. If I'm right about grain aliasing, it would be most effective with higher ISO slide film. I don't think I'll try this myself, since I have no need to scan old slides right now, and I haven't made any new ones in at least two decades.

 

So yeah, do your own testing ideally from a rental, then you'll know....

 

Great! Thank you. Now I'll rent a Sony and A/B. My interest around pixel shift was exactly about handling grain structure so this is encouraging.

[bjorn_solberg1]

This may be slightly off topic, but here is a very good example of a pixel shift scan done with a Pentax K-1. The Pentax K-1 has a 36mp Sony sensor, and some people (Lloyd Chambers: diglloyd blog: Pentax K1: Impressive Innovation Unmatched by CaNikon) claim that with pixel shift this camera can produce more than 50mp files.

 

Scroll down the page to member "Andrea K" and and take a look at his post. He scans Pentax 645 medium format film by taking four pixel shift shots and stich in Photoshop.

 

How do you scan your film? - Page 3 - PentaxForums.com

[babouphoto]

This may be slightly off topic, but here is a very good example of a pixel shift scan done with a Pentax K-1. The Pentax K-1 has a 36mp Sony sensor, and some people (Lloyd Chambers: diglloyd blog: Pentax K1: Impressive Innovation Unmatched by CaNikon) claim that with pixel shift this camera can produce more than 50mp files.

 

Scroll down the page to member "Andrea K" and and take a look at his post. He scans Pentax 645 medium format film by taking four pixel shift shots and stich in Photoshop.

 

How do you scan your film? - Page 3 - PentaxForums.com

 

Wow! That's actually really impressive. I'm now wondering if the same difference will be visible between 50mp MF and the new Sony!

[babouphoto]

I completely forgot that DPReview has their great test shots with comparisons. I just looked at the two side by side. The test allows you to see the Sony AR7IV in Pixel Shift mode and it is actually super interesting and very revealing.

 

The Sony has better contrast and looks sharper in general but not that much compared to the GFX 50. However, when pixel peeping, the GFX introduces a lot of color pixelation/artifacts which the Sony totally eliminates in Pixel Shift mode. It thus makes the Sony much more faithful and precise. Does not totally resolve my issue and I'm still not completely sure if the change would be worth it but it keeps pushing in Sony's favor. You can play with it here: Studio shot comparison: Digital Photography Review

 

Thanks!

[rodeo_joe1]

Scroll down the page to member "Andrea K" and and take a look at his post. He scans Pentax 645 medium format film by taking four pixel shift shots and stich in Photoshop.

Yes, that mosaic shot looks quite impressive, but there's still no sign of resolution down to the dye-cloud level. Making any assessment of a reduction in grain aliasing impossible.

 

There's also no 'before' shot to compare the pixel-shifted image to. So we still don't know if there's any real advantage to pixel shifting, and at what reproduction scale any advantage would reveal itself.

 

If it's only relevant at an RR where no further image detail, as opposed to grain detail, can be seen; then as far as I'm concerned the technique would have little to no practical use.

 

Many current cameras have a photosite/pixel spacing in the region of 4 microns anyway. That's without pixel shifting. This is also around the diameter of an individual dye cloud in a 100 ISO colour emulsion, and certainly smaller than the cluster size that happens with dye-cloud clumping. Such a 'scan' would result in a digital file of 196 megapixels from a 6x6cm film frame. And require a copying lens working at a ridiculously large aperture for diffraction not to completely mask the pixel 'resolution'.

 

So I remain extremely skeptical about the advantage of pixel shifting for this application.

Edited December 1, 2019 by rodeo_joe|1

[babouphoto]

Yes, that mosaic shot looks quite impressive, but there's still no sign of resolution down to the dye-cloud level. Making any assessment of a reduction in grain aliasing impossible.

 

There's also no 'before' shot to compare the pixel-shifted image to. So we still don't know if there's any real advantage to pixel shifting, and at what reproduction scale any advantage would reveal itself.

 

If it's only relevant at an RR where no further image detail, as opposed to grain detail, can be seen; then as far as I'm concerned the technique would have little to no practical use.

 

Many current cameras have a photosite/pixel spacing in the region of 4 microns anyway. That's without pixel shifting. This is also around the diameter of an individual dye cloud in a 100 ISO colour emulsion, and certainly smaller than the cluster size that happens with dye-cloud clumping. Such a 'scan' would result in a digital file of 196 megapixels from a 6x6cm film frame. And require a copying lens working at a ridiculously large aperture for diffraction not to completely mask the pixel 'resolution'.

 

So I remain extremely skeptical about the advantage of pixel shifting for this application.

 

Actually the DP Review studio test does allow to see before and after pixel shifting. I'm still with you about the actual benefits for negative scanning and going to rent the Sony camera today or next weekend to see for myself. However, the DP Review studio setup is really worth analyzing, especially the engravings which show done read aliasing in the GFX 50 shots which are not there in the Sony and GFX 100 shots...

[digitaldog]

However, when pixel peeping, the GFX introduces a lot of color pixelation/artifacts which the Sony totally eliminates in Pixel Shift mode.

Exactly and due to the ability of the Sony to capture true, trilinear RGB data vs. interpolated RGB data which I've already outlined. It's absolutely NOT a trick, it's not Mumbo Jumbo as anyone with actual experience shooting say a scanning back (Betterlight) vs. a single shot digital back (Leaf Cantare) that produces interpolated color would understand:

http://digitaldog.net/files/Filmvsdigital.pdf

Edited December 1, 2019 by digitaldog

[digitaldog]

Yes, that mosaic shot looks quite impressive, but there's still no sign of resolution down to the dye-cloud level. Making any assessment of a reduction in grain aliasing impossible.

Soon, there should be, once someone actually makes proper testing of both processes. Then the guessing and assumptions can finally end.

So I remain extremely skeptical about the advantage of pixel shifting for this application.

Skepticism based on assumptions. You might be right, I'm extremely skeptical it will NOT show a visible improvement and that's based on assumptions too.

When can we finally stop assuming and get to the facts? When the OP or someone else produces an actual TEST and does so with good testing parameters.

Edited December 1, 2019 by digitaldog

[bjorn_solberg1]

Yes, that mosaic shot looks quite impressive, but there's still no sign of resolution down to the dye-cloud level. Making any assessment of a reduction in grain aliasing impossible.

 

There's also no 'before' shot to compare the pixel-shifted image to. So we still don't know if there's any real advantage to pixel shifting, and at what reproduction scale any advantage would reveal itself.

 

If it's only relevant at an RR where no further image detail, as opposed to grain detail, can be seen; then as far as I'm concerned the technique would have little to no practical use.

 

Many current cameras have a photosite/pixel spacing in the region of 4 microns anyway. That's without pixel shifting. This is also around the diameter of an individual dye cloud in a 100 ISO colour emulsion, and certainly smaller than the cluster size that happens with dye-cloud clumping. Such a 'scan' would result in a digital file of 196 megapixels from a 6x6cm film frame. And require a copying lens working at a ridiculously large aperture for diffraction not to completely mask the pixel 'resolution'.

 

So I remain extremely skeptical about the advantage of pixel shifting for this application.

 

Did you download the 97mb jpeg of the scanned image? It gives a rather good impression of a Pentax K-1 pixel shift stiched scan.

 

Yes, this was a comparison between the Epson 4990 flatbed scanner and the Pentax K-1 pixel shifted. I have unfortunately not encountered any comparions between with pixel shift vs without pixel shift.

[bjorn_solberg1]

Once more slightly off topic: Here is a comparison between Pentax K-1 with pixel shift vs no pixel shift, but in a landscape context:

 

Pentax Pixel Shift and Motion

[Ed_Ingold]

This is a test, scanning a Kodachrome 25 slide with and without pixel-shifing, using a Sony A7Riii and Nikon 55/2.8 Macro lens. The upper panel is without. The difference is small, even at the dye cloud level. The image is obviously limited by the camera (Leica M2 + Summicron 50/2, probably at f/5.6) and medium (Kodachrome). Is it worth the extra frames and processing? Probably not. The overall image looks reasonably crisp, but pixel peeping was not in fashion in 1965.

 

Because the pixels are exceedingly small, I made this composite from 250x167 pixel samples from the original RAW images in Photoshop, then resampled to enlarge the composite by 4x.

 

Dye clouds do not have sharp boundaries, which limits the effect of pixel shifting, or better resolution in general. Pixel shifting has the greatest effect for landscapes and architecture, subject to resolution of the lens.

 

Edited December 1, 2019 by Ed_Ingold

[Ed_Ingold]

In this example, I compare a single frame rendition (top) with a pixel-shifted version (bottom) of the same scene. Again I took a 250 pixel wide sample of each, combined them into a composite, and resampled the results 4x. The shot was taken with a Sony AxRiii + Loxia 50/2. The conversion was made with PixelShift2DNG, and the samples were taken from the AWG and DNG raw images respectively, without sharpening. The difference is significant, but only with a side by side comparison. The equivalent magnification, re an 8x10" print, is 12x.

 

The Loxia 50 is not the sharpest lens available, but it is consistently sharp from corner to corner, with good color contrast. My sharpest lens (at landscape distance) is probably a Batis 135/2.8 APO. I didn't have it at the time, so a comparison must wait for another day.

 

It is noteworthy that there is no visible color aliasing in either image, even in the neutral toned areas.