A7rm4 workflow compo

[ray_parkhurst]

Since I bought the A7Rm4, I have only been shooting in jpg for a few reasons. First, I found that the jpg quality, given judicious camera settings for the jpg processor, is quite good. Second, the workflow for stacking and stitching with pixel shifting is daunting, and strained both my patience and my computer. And finally, the processors I've tried over-sharpen the pixel-shifted images to an unacceptable degree

At some point recently, Sony did an update to their software to allow more flexibility in sharpening composited files. I discovered this after upgrading to a new computer and installing the latest version of their software. This encouraged me to make a comparison between the jpg output I've been using versus a 4-shot pixel-shift composite. Of course one advantage to the composite is you can output in TIF format, and stacks in TIF retain more detail than do similar stacks in jpg, but for now I thought I'd just compare a single image.

Sony A7R4, Nikon 95mm Printing-Nikkor, nominal f3.3, mag 1.8x.

Here are single images showing the overall output, greatly reduced in size, to show the overall field of view and the similarity between the two processes:

jpg

pixel-shift -> tif -> jpg

And here are two crops, presented at 400%, to show the results of the two workflows:

jpg

pixel-shift ->tif -> jpg

The over-sharpening artifacts are gone with the new software revision, and as expected, the PS image has richer color detail, but it's not as strong as expected.

It may be that the EA9.2 of these images is limiting the improvement in pixel-shifting.

[rjlittlefield]

Also interesting is that the first crop gives me a strong sense that it exaggerates features at 45 degrees to the axes.

I see no trace of that in the second crop.

Do you see the same thing, and if so does it appear in other normal jpg's also?

--Rik

[ray_parkhurst]

I do see what you are saying, a general emphasis of upper right to lower left features. Very interesting, and I hadn't seen it before, or at least hadn't noticed it. I'll check to see if it's common across the image.

[Scarodactyl]

That seems to happen woth some sharpening algorithms, at least on some subjects.

[Beatsy]

I use A7riv and 4-shot pixel shift for photomicrography. I haven't downloaded the latest Sony Imaging Edge s/w yet - I'm not interested in the cloud features. But I will try it if there's improvement in pixel shift output to tiff.

Meanwhile, I found a couple of things that improve the pixel shift results with the (now) previous version - or an alternative.

First, I don't use "shortest" for the delay between pixel shift shots anymore. Using 1 second instead, I got far fewer "checkerboard" artefacts after processing the shifted shots through Imaging Edge (as in none compared to occasional small patches with the shortest setting). Similarly, I often perceived a sort of "shift" look in the image at 100%-200% zoom - somewhat like Rik described (and I see it in that image too). That went away with the 1-second delay too.

Second, I recently switched from using Imaging Edge and back to PixelShift2DNG (free). It consistently gives better results overall IMO. The combined image just looks cleaner, less "fizzy" - somehow.

All very subjective - but I'm definitely happier with the outcomes now - even if it is some kind of placebo effect. Might be worth a try if you're not already doing these things.

[ray_parkhurst]

I use A7riv and 4-shot pixel shift for photomicrography. I haven't downloaded the latest Sony Imaging Edge s/w yet - I'm not interested in the cloud features. But I will try it if there's improvement in pixel shift output to tiff.

Meanwhile, I found a couple of things that improve the pixel shift results with the (now) previous version - or an alternative.

First, I don't use "shortest" for the delay between pixel shift shots anymore. Using 1 second instead, I got far fewer "checkerboard" artefacts after processing the shifted shots through Imaging Edge (as in none compared to occasional small patches with the shortest setting). Similarly, I often perceived a sort of "shift" look in the image at 100%-200% zoom - somewhat like Rik described (and I see it in that image too). That went away with the 1-second delay too.

Second, I recently switched from using Imaging Edge and back to PixelShift2DNG (free). It consistently gives better results overall IMO. The combined image just looks cleaner, less "fizzy" - somehow.

All very subjective - but I'm definitely happier with the outcomes now - even if it is some kind of placebo effect. Might be worth a try if you're not already doing these things.

Going to a longer delay makes a lot of sense. I've been using shortest since it's the default, and had not experimented with longer delays. Perhaps that small shift is enough to impart a vibration that needs to settle before the next shot...

I tried PS2DNG but it made the workflow impossible on my computer. The new computer probably is OK with it, but the main reason I wanted to use it was to tame the over-sharpening, and the latest IE seems to do that OK. That "fizzy" look is gone IMO, at least with the processing I used.

Interesting about the offset changing. That offset is present in every composite I've made. I believed it was an artefact of the process, ie that the 4-shot composites are naturally offset from single images due to the shift. Not sure why it would go away, but also not really sure why it is there in the first place.

I will try the suggestions and see what happens. Thanks.

[Beatsy]

...
Going to a longer delay makes a lot of sense. I've been using shortest since it's the default, and had not experimented with longer delays. Perhaps that small shift is enough to impart a vibration that needs to settle before the next shot...

My hypothesis too. I never saw this on my ICM 405 where the camera and adapter are directly attached onto a T-mount thread that protrudes from the cast-iron frame. Similarly, no problems on my macro rigs where camera and tubes are all firmly held in place. But on the new (to me) BX61 it's a typical trinocular arrangement, with an adapter held in by a single grub screw pressing against a dovetail mount. There are some degrees of freedom to vibrate. Tiny amount but enough, it seems. The sensor mechanism is quite a chunk to be throwing around at speed - so it's at least feasible that it could set up a short term vibration.

I tried PS2DNG but it made the workflow impossible on my computer. The new computer probably is OK with it, but the main reason I wanted to use it was to tame the over-sharpening, and the latest IE seems to do that OK. That "fizzy" look is gone IMO, at least with the processing I used.

Sounds like the Sony s/w might be fixed. I'll download and try it when I get time.

Interesting about the offset changing. That offset is present in every composite I've made. I believed it was an artefact of the process, ie that the 4-shot composites are naturally offset from single images due to the shift. Not sure why it would go away, but also not really sure why it is there in the first place.

Vibration in a preferred direction could explain that too IMO.

I will try the suggestions and see what happens. Thanks.

Let us know if it changes anything. Thanks.