Sony Camera Upgrades - Color and Pixel Shift

[silvergelatin]

I am gradually upgrading pieces of my scanning system (ordered a Scanlight and will likely trade in my 70mm Macro Art lens for the 105), and am considering trading in my A7RII for either the III or IV. The main reason is to have pixel shift available to get better color, shadow detail, and reduced noise. Most threads I have read seem to focus on spatial resolution, but I’m more concerned with the other benefits.

Obviously the IV is the more serious upgrade on paper, since the sensor is newer and higher res, but does it actually provide better color and noise performance compared to the older models? For example, if I ran a four shot pixel shift with the III and the IV, would there be any advantage to the IV other than spatial resolution?

 

[koraks]

Pixel shift has come up a couple of times in this context here on Photrio, but I don't recall anyone reporting any significant practical benefit.

 

[silvergelatin]

Pixel shift has come up a couple of times in this context here on Photrio, but I don't recall anyone reporting any significant practical benefit.

Those threads seem to focus on spatial resolution and not the other aspects.

 

[koraks]

Okay, gotcha.

 

[Ardpatrick]

I can't answer your question, because I haven't tried it. But I've been reading quite a bit online about camera scanning and pixel shift myself. My guess is the difference between the III and the IV would be minor, beyond the colour benefit they both derive from pixel shift in general. There does seem to be fairly wide consensus about the benefits of pixel shift for colour, but the evidence around spatial resolution is less conclusive. Something I haven't seen discussed widely for example, which would be a consideration for your new set-up, is the resolving capabilities of the lens you're using. Can the lens resolve anything close to the 240 megapixel file the AR7 IV produces?

 

[loccdor]

There really shouldn't be much noise to speak of from any version of the camera at the low ISO you would use for film digitization.

Color reproduction improves by using pixel shift vs. not. How much, I haven't tested, but others will have results online you can look at.

The IV version of the camera has a half-pixel shift 16 shot mode for 230 MP files. I've never tried such a mode, my Pentax K-1 (first version) has 4 shot mode. I'd expect diminishing returns and huge files. If the resolution is truly there, then expect to struggle a lot with film flatness, parallelism, diffraction, and focus, in order to achieve it. It would also slow down the rate at which you can digitize a roll.

With film digitization, you'd need an extremely contrasty negative/positive for the shadow detail of your digitizing camera to come into play. I don't think this will matter much for 99% of circumstances.

Pixel shift was the deciding factor when comparing a 50MP camera without it to a 36MP with it in the same price bracket. Your III and IV cameras seem to be pretty close in price and one has quite a few more megapixels. If you're primarily a 35mm shooter, I'd personally go for something with pixel shift cheaper than both of them, unless you consistently use very high resolution film. If you shoot tons of medium format, maybe it will be useful for you.

 

[Ardpatrick]

To add - if you’re after improved colour the light source will be really important. I use flash (as opposed to continuous light sources) and I know that some but not all cameras that do pixel shift can work with flash.

Whatever about the relative CRI merits of continuous vs flash, at least I know I’m not suffering with camera shake whilst shooting very highly magnified images when using flash. For critical sharpness flash is hard to beat.

 

[4season]

I very much like the pixel-shift feature, not just for the added resolution, but for the way it can eliminate aliasing artifacts. But to date, it hasn't seemed particularly useful for scanning film. Maybe if I were dealing with microfilm or microfiche, or wished to digitize high-res halftones?

Negatives don't seem to challenge the digital camera at all with regard to colors or dynamic range. Likely that 8-bit JPEG would provide more than adequate coverage. Nevertheless, scanning can be rewarding, as there can be a good deal of useful detail in shadows and highlights, albeit tonally compressed.

Slides can be more challenging in terms of dynamic range, and to date, my efforts to extract additional shadow and highlight detail have been met with very modest success: There simply doesn't seem to be much data there, and what little there is is often heavily compressed.

Fujifilm Instax prints have been a pleasant surprise though: There can be a decent amount of tonality in the highlights which isn't always immediately apparent when viewing the original print.

 

[silvergelatin]

To add - if you’re after improved colour the light source will be really important. I use flash (as opposed to continuous light sources) and I know that some but not all cameras that do pixel shift can work with flash.

Whatever about the relative CRI merits of continuous vs flash, at least I know I’m not suffering with camera shake whilst shooting very highly magnified images when using flash. For critical sharpness flash is hard to beat.

I am currently using flash for color negs, but I'm expecting that the RGB Scanlight will be able to replace that once it gets here. I do find flash to be a big step up for color compared to a typical white LED panel, though.

 

[silvergelatin]

There really shouldn't be much noise to speak of from any version of the camera at the low ISO you would use for film digitization.

Color reproduction improves by using pixel shift vs. not. How much, I haven't tested, but others will have results online you can look at.

The IV version of the camera has a half-pixel shift 16 shot mode for 230 MP files. I've never tried such a mode, my Pentax K-1 (first version) has 4 shot mode. I'd expect diminishing returns and huge files. If the resolution is truly there, then expect to struggle a lot with film flatness, parallelism, diffraction, and focus, in order to achieve it. It would also slow down the rate at which you can digitize a roll.

With film digitization, you'd need an extremely contrasty negative/positive for the shadow detail of your digitizing camera to come into play. I don't think this will matter much for 99% of circumstances.

Pixel shift was the deciding factor when comparing a 50MP camera without it to a 36MP with it in the same price bracket. Your III and IV cameras seem to be pretty close in price and one has quite a few more megapixels. If you're primarily a 35mm shooter, I'd personally go for something with pixel shift cheaper than both of them, unless you consistently use very high resolution film. If you shoot tons of medium format, maybe it will be useful for you.

I do find some minor shadow noise on contrasty color negs quite often, and that is one of the reasons I am upgrading. Other than that, I do realize I am just squeezing out the last 1-5% of potential quality here. After trying different light sources and settling on the Sigma lens, the only other way to get even better would be to look into exotic process lenses, but that opens another can of worms.

 

[Ardpatrick]

I am currently using flash for color negs, but I'm expecting that the RGB Scanlight will be able to replace that once it gets here. I do find flash to be a big step up for color compared to a typical white LED panel, though.

I guess putting my above posts together, what I'm pointing to is your choice of camera, and to check whether either / both can do pixel shift with flash exposure. As I said above, quite a few recent mirrorless cameras can do pixel shift, but not with flash for each exposure. If your new light source works out, great! You wouldn't need a camera with that flash capacity. But flash is a very nice light source for this kind of work, and frankly hard to beat. I'm sure those higher end Sony's likely do have the capability anyway.

 

[silvergelatin]

I guess putting my above posts together, what I'm pointing to is your choice of camera, and to check whether either / both can do pixel shift with flash exposure. As I said above, quite a few recent mirrorless cameras can do pixel shift, but not with flash for each exposure. If your new light source works out, great! You wouldn't need a camera with that flash capacity. But flash is a very nice light source for this kind of work, and frankly hard to beat. I'm sure those higher end Sony's likely do have the capability anyway.

I was deciding between grabbing a small monolight to have the convenience of a modeling lamp, or trying the RGB light. Consensus seems to be that the RGB should be slightly better or at least as good for color, but we will see.

 

[loccdor]

I do find some minor shadow noise on contrasty color negs quite often, and that is one of the reasons I am upgrading. Other than that, I do realize I am just squeezing out the last 1-5% of potential quality here. After trying different light sources and settling on the Sigma lens, the only other way to get even better would be to look into exotic process lenses, but that opens another can of worms.

I wonder if you'd improve the color neg results by using an 80A filter to bring the R/G/B channels closer together on the histogram. I plan on trying this soon.

 

[silvergelatin]

I wonder if you'd improve the color neg results by using an 80A filter to bring the R/G/B channels closer together on the histogram. I plan on trying this soon.

I believe the RGB Scanlight may already be balanced to neutralize the mask, so I'm hoping it has a positive effect on noise.

 

[MattKing]

I believe the RGB Scanlight may already be balanced to neutralize the mask, so I'm hoping it has a positive effect on noise.

So you are expecting it to add a bunch of cyan, in order to remove the effect of the calibrated to red mask?

 

[silvergelatin]

So you are expecting it to add a bunch of cyan, in order to remove the effect of the calibrated to red mask?

I believe another user implied that in another thread. Maybe a different forum. If not, the RGB levels are adjustable via software, so that could be done.

 

[mshchem]

So RGB scan lighting? Makes 3 separate exposures?

 

[silvergelatin]

So RGB scan lighting? Makes 3 separate exposures?

It's a combined RGB light, not separate.

A Better Light Source For Scanning Color Negative Film

jackw01.github.io

 

[mshchem]

It's a combined RGB light, not separate.

A Better Light Source For Scanning Color Negative Film

jackw01.github.io

So I wonder how this compares to a hot light? I use old school colorheads for making RA4 prints. I wonder how the spectra compare? Not sure I'm buying into this, but I'm easily confused.

 

[silvergelatin]

So I wonder how this compares to a hot light? I use old school colorheads for making RA4 prints. I wonder how the spectra compare? Not sure I'm buying into this, but I'm easily confused.

Me too . I've seen a few threads around that show conversions made with the Scanlight vs normal high CRI panels, and the Scanlight was significantly better. My understanding is that by narrowing the color bands to line up better with the camera sensor, it should improve channel separation. Looking forward to trying it.

 

[mshchem]

I wonder how diffuse sunlight for a light source would work?

Also, how narrow is the spectra of RGB LEDs? I mean this isn't element emission spectra??

Heiland uses individual RGB LEDs for the fancy enlarger light sources, is this a similar thing?

 

[silvergelatin]

I wonder how diffuse sunlight for a light source would work?

Also, how narrow is the spectra of RGB LEDs? I mean this isn't element emission spectra??

Heiland uses individual RGB LEDs for the fancy enlarger light sources, is this a similar thing?

Might want to email Jack about the spectral specifics.

As far as sunlight goes, it would be extremely inconvenient. Flash or old Solux bulbs should provide most of the benefit, I would think.