Monochrome Camera / Techniques?

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

Lou Jost
Posts: 5947
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Monochrome Camera / Techniques?

Post by Lou Jost »

Steve, most cameras do not export four shifted RAWs, so that was why I didn't think that is what you meant. I think Sony may be the only one; you mentioned Pentax but the K-1 (which I have) does the combining in-camera, like Olympus and Panasonic and all others that I can think of. None of these allow manipulation of the individual shifted images.

jmc
Posts: 238
Joined: Fri Apr 17, 2020 3:14 am

Re: Monochrome Camera / Techniques?

Post by jmc »

With regards to sensor sensitivity measurements as a function of wavelength, I've assessed a few cameras;

Sony A7III here - https://jmcscientificconsulting.com/pro ... beginning/

Canon EOS 5DSR here - https://jmcscientificconsulting.com/tes ... er-filter/

If you have access to the Royal Photographic Societies Imaging Science Journal, I've written an article on it here - https://www.tandfonline.com/doi/abs/10. ... ode=yims20
Jonathan Crowther

Steve S
Posts: 84
Joined: Fri May 10, 2013 1:52 pm
Location: Southern Arizona, USA

Re: Monochrome Camera / Techniques?

Post by Steve S »

Lou Jost wrote:
Sun Sep 06, 2020 4:21 pm
Steve, most cameras do not export four shifted RAWs, so that was why I didn't think that is what you meant. I think Sony may be the only one; you mentioned Pentax but the K-1 (which I have) does the combining in-camera, like Olympus and Panasonic and all others that I can think of. None of these allow manipulation of the individual shifted images.
Lou, I'll happily stand corrected except for: https://www.fastrawviewer.com/PixelShift2DNG. Pentax must do it both ways.

Not being a Pentax owner since about 1963, I'm not an authority, but it is generally safe to trust Mr Borg, almost as reliable as lyin' eyes. Sony has a 16- as well as 4- shot shift, an awful lot of files.

Lou Jost
Posts: 5947
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Monochrome Camera / Techniques?

Post by Lou Jost »

We're both right.

"When it launched its crop-sensor flagship camera, the Pentax K-3 II, back in April, 2015, Ricoh debuted a brand-new feature that was hugely useful to anyone wanting the maximum per-pixel sharpness and dealing primarily with relatively static subjects. Dubbed Pixel Shift Resolution (or in some markets, Real Resolution), this function captured multiple images and combined them in-camera to create a single shot with higher resolution."
https://www.imaging-resource.com/PRODS/ ... R_MODE.HTM

"In short, the K-1 will take four exposures in rapid succession (usually taking a second or two in our tests) and will then process those four images into a single image."
https://digitalrev.com/2016/05/17/is-pe ... rth-using/

"When shooting is performed, captured four images are displayed in turn, and then they are merged into a single image."
K-1 Manual

I do not recall ever processing the four images externally.

But reading the instructions you linked to, I see the output files are called "4-shot PEF/DNG containers" so I guess this program gets into the innards of the (single) container file. Pentax's and third parties' raw processors may be doing this "under the hood".

I have just taken a normal image and a pixel-shifted image withy my K1. It savs only one jpg and only one RAW for each image. The normal RAW is 46209kb and the shifted one is 173805km, fairly close to 4x the individual RAW. So there must be four independent RAWs in that file, which can be extracted and manipulated separately. That's a nice option to have, though I do not recall noticing over-sharpening of the RAW shifted files from the K1.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

Lou Jost wrote:
Mon Sep 07, 2020 11:53 am
But reading the instructions you linked to, I see the output files are called "4-shot PEF/DNG containers" so I guess this program gets into the innards of the (single) container file. Pentax's and third parties' raw processors may be doing this "under the hood".

I have just taken a normal image and a pixel-shifted image withy my K1. It savs only one jpg and only one RAW for each image. The normal RAW is 46209kb and the shifted one is 173805km, fairly close to 4x the individual RAW. So there must be four independent RAWs in that file, which can be extracted and manipulated separately. That's a nice option to have, though I do not recall noticing over-sharpening of the RAW shifted files from the K1.
When I evalusted the K3-II a few years ago, it was the first time I had seen the unnatural-looking sharpening in pixel-shift images. There were no controls to improve the IQ, so I sent it back. Are you guys saying that PS2DNG will re-compose the composite image, and will do so without excessive sharpening? Same thing with the K1? That would open up another potential option for me.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

Steve S wrote:
Sun Sep 06, 2020 10:15 am
Ray, you had expressed concern about pixel-shift processing software oversharpening. I was sure that Iliah Borg's PixelShift2DNG (still a free beta) would not, but, pennies being a readily available subject, I tried it. An A7RM4 was mounted on a Wild M400 with a ringlight, so that's what I used. With the ringlight undiffused there was a lot of that funky speckle false color that we now kinda sorta understand; and, with a lot of diffusion, very little, as we would expect. In both cases, the single-frame Bayer images are very, very similar to the shifted-and-merged image. Except in an A-B situation you couldn't distinguish them. In both cases, the subjective impression was that the single frame was slightly crisper, or, put another way, that the merged image could take (and needed) more sharpening.

Conclusion: here, 4-pixel shifting does no good and does no harm. The workflow doesn't amount to much: the multi-pixel shots require a single press of the release, so you could run an automated stack in just the normal way. With all the images in a folder, just point PixelShift2DNG to it, and click "Analyze + Convert All", and Iliah's little utility magically generates merged DNG's. So easy it's too bad there's no benefit in this situation. The qualification is necessary because the image is diffraction limited by the lens, not pixel-pitch limited by the sensor. (The Wild ApoZoom is a lot of fun, but the NA doesn't compete with some of today's optics -- it was intended for film.)
Steve...thanks for the good info. I was not aware that PS2DNG could batch convert, though it makes good sense it does. I only tried it once on a single 4-shot composite I had saved from my A7Rm4 evaluation, and had assumed I'd need to do an extremely tedious workflow. So let's say I do a 3x4 panorama, each tile with 20-shot stacks, total 240 shots that need to be kept in proper order. Let's say I did a 4-shot PS composite for each, I'd be looking at 960 images to dump into PS2DNG. I assume I'd get 240 .dng composite shots, correct? I can't stack .dng files so is there a good batch converter to do .dng to .tiff? If so, then the workflow might not be too onerous.

Regarding the resolution issue, I'd be using a lens with large enough aperture to make the 4 (or even 16) image PS worthwhile. With 16-images, I'd be looking at 3840 images, so batch workflow is ultra critical...Ray

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

dolmadis wrote:
Sat Sep 05, 2020 12:05 pm
Thanks Ray.

An intricate lighting design in every way which has rewarded you with fine imaging.

I have learnt a very great deal visualising the set up and must thank you very much for sharing given that it is the product of many hours work.
You are welcome! I developed the lighting method overy several years of trial and error. The lighting angles are very critical to the overall look of the image.

As an example, I was encouraged to re-test the Coolscan-4/5 Scanner-Nikkor ED 7-element lens after Robert's very positive recent review. He tested at 1.4x, so that in reverse it should be optimum at 0.7x. Indeed the resolution is excellent, but due to the short working distance, lighting is sub-optimum. I believe I could improve the lighting to be similar to the 60WA image by using a much smaller ringlight, but not with my standard ringlight system. See the images from the 60WA and the SNED4 below. To be sure the SNED4 image is not bad, it just falls short compared with the 60WA due to the available lighting.

60WA
IMG_0001 - Copy_2.JPG
SNED4
IMG_4056 - Copy_2.JPG

Lou Jost
Posts: 5947
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Monochrome Camera / Techniques?

Post by Lou Jost »

Are you guys saying that PS2DNG will re-compose the composite image, and will do so without excessive sharpening?
That's what I would now conclude from all this. And it would do so without your having to save the 4 individual shifted images; they get saved into a single file.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

Lou Jost wrote:
Mon Sep 07, 2020 2:26 pm
Are you guys saying that PS2DNG will re-compose the composite image, and will do so without excessive sharpening?
That's what I would now conclude from all this. And it would do so without your having to save the 4 individual shifted images; they get saved into a single file.
OK, so now I may need to purchase (or re-purchase) a pixel-shifting camera, but which one? I know how the A7Rm4 works, but have not owned a K1, S1R, or Olympus. I assume the A7Rm4 is the only one with Live View tethering capability, correct? Which has best 100% image quality?

Lou Jost
Posts: 5947
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Monochrome Camera / Techniques?

Post by Lou Jost »

I would not recommend the K1. It is not mirrorless, so your lens choices are limited (though you can convert the mount to Nikon, as I did) and you will rapidly accumulate lots of shutter actuations when stacking with stitching. Strangely, the shutter fires even when using silent electronic shutter; it just fires after the exposure. Stupidest thing I've ever seen.

If you want 4-shot pixel shifting (which can be free of artifacts, because you are just measuring RGB at all pixels, not using calculations to increase the pixel resolution) and if the Pentax is eliminated, then you are limited to the Sony. Olympus and Panasonic give you 8-shot shifting.

Steve S
Posts: 84
Joined: Fri May 10, 2013 1:52 pm
Location: Southern Arizona, USA

Re: Monochrome Camera / Techniques?

Post by Steve S »

Ray, as to your concern about converting DNG to TIFF, these merged pixel-shifted DNGs are not like everyday, garden-variety Bayer DNGs: they have full color information at each pixel, and do not require demosiacking. They are very much like TIFFs except that the color information is not transformed into a particular color space. That's the kind of DNG that Helicon focus produces. I only have this on the authority of Jim Kasson, but he's rarely wrong. By the way, I think this fear of added sharpening comes from fooling with JPGs. Most responsible raw workflows don't promiscuously add sharpening. (And we must remember that sharpening, like ethanol, can work wonders in moderation.)

As to cameras, think about lens coverage. If you are not going to use lenses that cover more than m43, then the Oly might be an excellent choice, but otherwise a FF camera would be better. I encourage everyone to get one of them high-end Pannys: I'm curious about them, but so heavily into E- and GF- mount that I must experience them only vicariously.

Please apply the modifier "I think" to each and every predication above.

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

Steve S wrote:
Tue Sep 08, 2020 7:18 am
Ray, as to your concern about converting DNG to TIFF, these merged pixel-shifted DNGs are not like everyday, garden-variety Bayer DNGs: they have full color information at each pixel, and do not require demosiacking. They are very much like TIFFs except that the color information is not transformed into a particular color space. That's the kind of DNG that Helicon focus produces. I only have this on the authority of Jim Kasson, but he's rarely wrong. By the way, I think this fear of added sharpening comes from fooling with JPGs. Most responsible raw workflows don't promiscuously add sharpening. (And we must remember that sharpening, like ethanol, can work wonders in moderation.)

As to cameras, think about lens coverage. If you are not going to use lenses that cover more than m43, then the Oly might be an excellent choice, but otherwise a FF camera would be better. I encourage everyone to get one of them high-end Pannys: I'm curious about them, but so heavily into E- and GF- mount that I must experience them only vicariously.

Please apply the modifier "I think" to each and every predication above.
Steve...I would not mind working with DNG but according to Helicon they don't support stacking DNGs. My normal workflow is shoot raw, convert to tiff, stack and tile tiff, output jpg. With the A7Rm4 I suppose it would be shoot raw, batch convert to dng, somehow convert to tiff, stack and tile tiff, output to jpg.

The oversharpening I complained about was from Imaging Edge, which outputs jpg only I believe, but that may be different with later versions. If they have fixed their problem then I'd just go with that. I view anything else as a workaround until Sony fixes the problems with their software. I did try PS2DNG and it gave good IQ from a 4-image pixel-shift composite, so I know I can go that way.

I'm interested in FF only to increase coverage so that I can minimize the number of tiles in my panoramas. Right now with the 95PN at 2x on APS-C I can do a 2x3 subject-stitched panorama with good results. My goal is to use the Rayfact 3.5x on FF with 2x3 sensor-stitched panorama.

Steve S
Posts: 84
Joined: Fri May 10, 2013 1:52 pm
Location: Southern Arizona, USA

Re: Monochrome Camera / Techniques?

Post by Steve S »

Ray,

1) Helicon Focus can use Adobe Raw as its processing engine; Adobe Raw of course supports numerous DNG formats from cameras/phones. I don't have a license, trying to stay loyal to Rik and in my comparisons during a recent trial of Helicon I saw no reason to buy their license, so I can't test this, but I'd be surprised if you couldn't just pass intermediate DNG's to Helicon for stacking. Have you actually tried this?

2) I use Imaging Edge only for tethering, but I tried opening those four penny-shift ARW's in IE Viewer, which offered to merge them; when it then automatically opened the merged image in IE Edit (oi vey, why three programs??), IE Edit was happy to save the result as an ARQ, which I imported into LR and compared to the DNG that came from PixelShift2DNG. I can't see any real difference, and I didn't fool around with the sharpening settings in IE Edit.

3) Expressed as a square format (since most coins are circular), what are you looking for in the way of final pixel dimensions of the reference image? What is the effective aperture of that big lens at the magnification you are using (wondering what pixel density you actually need)? Finally (just curiosity), what software are you using for the lateral merging?

ray_parkhurst
Posts: 3416
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by ray_parkhurst »

Steve S wrote:
Tue Sep 08, 2020 10:49 am
Ray,

1) Helicon Focus can use Adobe Raw as its processing engine; Adobe Raw of course supports numerous DNG formats from cameras/phones. I don't have a license, trying to stay loyal to Rik and in my comparisons during a recent trial of Helicon I saw no reason to buy their license, so I can't test this, but I'd be surprised if you couldn't just pass intermediate DNG's to Helicon for stacking. Have you actually tried this?

2) I use Imaging Edge only for tethering, but I tried opening those four penny-shift ARW's in IE Viewer, which offered to merge them; when it then automatically opened the merged image in IE Edit (oi vey, why three programs??), IE Edit was happy to save the result as an ARQ, which I imported into LR and compared to the DNG that came from PixelShift2DNG. I can't see any real difference, and I didn't fool around with the sharpening settings in IE Edit.

3) Expressed as a square format (since most coins are circular), what are you looking for in the way of final pixel dimensions of the reference image? What is the effective aperture of that big lens at the magnification you are using (wondering what pixel density you actually need)? Finally (just curiosity), what software are you using for the lateral merging?
I was going by Helicon's list of supported file formats, which don't include DNGs. Where would I look to see how to work directly with DNGs?

I wrote up a mini-review of the A7R4. I'm amazed that this review was written less than a year ago:

http://www.photomacrography.net/forum/v ... r4#p255133

With the 95PN at 2x and 4.3um sensor pixels, my output images are 9.6kp square for one coin side (92MP), or 10.2x9.6kp for a side-by-side image. The HRT2i sensor is 5184x3456, so my tiling efficiency in each dimension is 92.5%. You can see an example side-by-side image on EasyZoom here:

https://easyzoom.com/image/157586/album/0/4?mode=manage

Going from APS-C to FF I need to increase magnfiication to ~3.25x. The A7Rm4 sensor is 9504x6336, or 83.3% more pixels on each side. Assuming I can maintain ~90% efficiency, the merged file will be ~17kp square, or 289MP.

The 95PN is operating at ~EA10. The Rayfact 3.5x will be operating at ~EA12. That should still be wide enough for pixel-shifting to make sense, at least 4-image composites.

I've been using Microsoft's old Image Composite Editor for the tiling. I see that the latest version of Helicon has a panorama stitcher, so I will probably try that. ICE is extremely finicky but gives an excellent result.

rjlittlefield
Site Admin
Posts: 23562
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Monochrome Camera / Techniques?

Post by rjlittlefield »

ray_parkhurst wrote:
Tue Sep 08, 2020 1:04 pm
I was going by Helicon's list of supported file formats, which don't include DNGs. Where would I look to see how to work directly with DNGs?
Try just dragging them in. I just now tested with Helicon's plug-in for Lightroom, doing an export as DNG (from sources in JPEG). Lightroom took a while to do the format conversion, then sure enough, I ended up in Helicon Focus with a bunch of .dng files showing in the list of Source images, and a bunch of 16-bit TIFF files accumulating in HF's cache folder. This looks like what I've seen before with other types of raw files -- that HF's workflow behind the scenes appears to be convert raw to 16-bit TIFF, stack the TIFFs, then if DNG output is selected, package the RGB-per-pixel output in a DNG wrapper as suggested above by Steve S and Jim Kasson. In my test, the DNG files fed into HF should be RGB-per-pixel, because of their origin as JPEG, so I would not expect it to have trouble with RGB-per-pixel DNG's from other sources such as PixelShift2DNG.

--Rik

Post Reply Previous topicNext topic