The lenses we use

[Lou Jost]

Having said all that, I can understand that if his goal is to cater to people who insist on judging a photo only at 100% resolution, his approach makes sense, though he also must have an implicit criterion about the size of the details that he wants to resolve.

Edit: Ray, I didn't see your last post when I posted this. I do sort of understand what you are after; I just think your two goals (more resolution, better 100% sharpness) are self-contradictory goals, given a fixed set of lenses.

[Justwalking]

Ray, interesting what if you take 4 shots let's say in accurate 2X [Pixel sensor size] acros 2 axis in both direction. Will be summary resolution also higher?

[Justwalking]

It's very frustrating that I keep telling you both what I'm after, but it isn't sinking in. I'm looking for BOTH more resolution (which will eventually come from upgrade to FF or possibly MF camera) AND an improvement in sharpness at 100% pixel level. Is that clear enough? If not, I don't know how to say it more clearly.

Ray, let's say the lens can give us object space very exellent as gradient in
luminosity. To achieve maximum accurate repeating on the sensor we need maximum density of pixels. To achieve maximum sharp we need as low steps to show this gradient as minimum. So we can choose only a compromisse between.

[ray_parkhurst]

Having said all that, I can understand that if his goal is to cater to people who insist on judging a photo only at 100% resolution, his approach makes sense, though he also must have an implicit criterion about the size of the details that he wants to resolve.

Edit: Ray, I didn't see your last post when I posted this. I do sort of understand what you are after; I just think your two goals (more resolution, better 100% sharpness) are self-contradictory goals, given a fixed set of lenses.

The two goals are pretty much independent, assuming the lens used is presenting more information to the sensor than it can capture. Lack of sharpness at 100% is a shortcoming of the sensor electronics, bayer demosaicing, any filters in the optical path, plus probably a handful of other factors I don't even know about. Resolution (again assuming the lens is capable) has mostly to do with number of sensor pixels.

You are correct about an implicit criterion for detail resolution, and this is the ultimate factor that will push me to FF. The "money shot" for variety attributions is the full Date and MM of the Lincoln Cent. Additionally, having the MM itself large enough for correct attribution and documentation is equally important. On my 18MP camera, I am (was) hampered on both ends...the 100% zoom image of a MM is not sharp enough for proper documentation/attribution, and is annoyingly small. Going to FF makes it bigger, but still not sharp enough for the work. With SR, I think the image is sharp enough even on APS-C, but still annoyingly small, which will improve with FF.

[ray_parkhurst]

Ray, interesting what if you take 4 shots let's say in accurate 2X [Pixel sensor size] acros 2 axis in both direction. Will be summary resolution also higher?

I would guess that you'd get a little better accuracy if you did 4 shots at exactly the right step, but that seems a tough thing to do because the optimum step size would vary with magnification. That's not a path I would care to go down.

[Justwalking]

On my 18MP camera, I am (was) hampered on both ends...the 100% zoom image of a MM is not sharp enough for proper documentation/attribution, and is annoyingly small. Going to FF makes it bigger, but still not sharp enough for the work. With SR, I think the image is sharp enough even on APS-C, but still annoyingly small, which will improve with FF.

"Sharp enough" in opposite to density of pixels for the FoV at100% crop.
Is this full frame 8MP 1/3" camera shot of 0.6 mm letter is not sharp enough
for proper documentation compared to SR D-letter? For large FoV with year sign on the coin you can also make several shots and stich them.

[ray_parkhurst]

On my 18MP camera, I am (was) hampered on both ends...the 100% zoom image of a MM is not sharp enough for proper documentation/attribution, and is annoyingly small. Going to FF makes it bigger, but still not sharp enough for the work. With SR, I think the image is sharp enough even on APS-C, but still annoyingly small, which will improve with FF.

"Sharp enough" in opposite to density of pixels for the FoV at100% crop.
Is this full frame 8MP 1/3" camera shot of 0.6 mm letter is not sharp enough
for proper documentation compared to SR D-letter?

That looks good from documentation perspective. This image is 1000x750, while your 8MP sensor is something like 3450x2300, correct? So this is about 1/3 the height of your sensor, meaning your total FOV is only 1.8mm high. I'm trying for about 20-21mm vertical FOV. Not quite the same ballpark.

edited to add:

I shot the same Cent using a 105PN at f3.3, 20 shot stacks, composite of 6 stacks for the SR. I am not sure if it's better to stack first, then SR composite the stacked images, or do SR first, then stack the results. The former is much less work so I went with it. The latter would need to be a whole lot improved to make sense, but given the result I don't think I need to go there.

This is an animation of 100% crops of the MintMark on the Cent, single stacked shot vs SR composite. It is acceptably sharp, but is a bit too small. Would be nice to be 2x bigger (MF) but 1.67x bigger (FF) seems like it would do nicely.

[Justwalking]

That looks good from documentation perspective. This image is 1000x750, while your 8MP sensor is something like 3450x2300, correct? So this is about 1/3 the height of your sensor, meaning your total FOV is only 1.8mm high. I'm trying for about 20-21mm vertical FOV. Not quite the same ballpark.

Ray, how you know what is resizing factor here? It is not 100% crop. This is full frame resized to 750. Sure 100% crop looks blurred.

[ray_parkhurst]

That looks good from documentation perspective. This image is 1000x750, while your 8MP sensor is something like 3450x2300, correct? So this is about 1/3 the height of your sensor, meaning your total FOV is only 1.8mm high. I'm trying for about 20-21mm vertical FOV. Not quite the same ballpark.

Ray, how you know what is resizing factor here? It is not 100% crop. This is full frame resized to 750. Sure 100% crop looks blurred.

So that is even farther afield than I was assuming. Your FOV is only ~1mm vertical, correct? I need 20-21mm.

[Justwalking]

So that is even farther afield than I was assuming. Your FOV is only ~1mm vertical, correct? I need 20-21mm.

Understand. But how much pixels need your for that if 1mm want 8MP for this resolution?
So you have 120 shots to see the coin (if your cover whole coin) I think 120 shots with stitch on 1/3" sensor can cover 20 mm coin with resolution like this and 100% crop will looks like this. Good reason to think about automation such stitch process.

[ray_parkhurst]

So that is even farther afield than I was assuming. Your FOV is only ~1mm vertical, correct? I need 20-21mm.

Understand. But how much pixels need your for that if 1mm want 8MP for this resolution?
So you have 120 shots to see the coin (if your cover whole coin) I think 120 shots with stitch on 1/3" sensor can cover 20 mm coin with resolution like this and 100% crop will looks like this. Good reason to think about automation such stitch process.

Stitching has always been an option, but I have not been interested in going there. It does not work all that well for coins with their specular highlights.

[rjlittlefield]

It's very frustrating that I keep telling you both what I'm after, but it isn't sinking in. I'm looking for BOTH more resolution (which will eventually come from upgrade to FF or possibly MF camera) AND an improvement in sharpness at 100% pixel level. Is that clear enough? If not, I don't know how to say it more clearly.

I sympathize with your frustration. Trust me, it has been really frustrating on this end too, and I've spent (wasted?) quite a few hours trying to puzzle it out.

But now I think I see where you're going: sharper pixels AND more of them.

I'm still not sure why you're willing to move the subject by subpixel amounts and run SR, versus moving it by larger amounts and stitching with telecentric optics, which would remove the specular highlights problem. But I digress.

By the way, to wax philosophical... I now think my confusion was caused by a difference in approach between you and me. If I can make progress toward a goal by two actions, one of which is simple and guaranteed to make a big improvement, while the other is difficult and uncertain, for sure I'll take the easy one first and lock in that gain while working on the hard one. You seem to be doing it the other way around. If you had just said that, then at least I would have been assured that you understood the issues and I could have skipped going to so much trouble to try explaining them.

I will watch the evolution of your SR efforts with interest.

--Rik

Edit: revised after seeing Ray's last post about stitching.

[ray_parkhurst]

It's very frustrating that I keep telling you both what I'm after, but it isn't sinking in. I'm looking for BOTH more resolution (which will eventually come from upgrade to FF or possibly MF camera) AND an improvement in sharpness at 100% pixel level. Is that clear enough? If not, I don't know how to say it more clearly.

I sympathize with your frustration. Trust me, it has been really frustrating on this end too, and I've spent (wasted?) quite a few hours trying to puzzle it out.

But now I think I see where you're going: sharper pixels AND more of them.

I'm still not sure why you're willing to move the subject by subpixel amounts and run SR, versus moving it by larger amounts and stitching, but I assume that will become clear eventually.

By the way, to wax philosophical... I now think my confusion was caused by a difference in approach between you and me. If I can make progress toward a goal by two actions, one of which is simple and guaranteed to make a big improvement, while the other is difficult and uncertain, for sure I'll take the easy one first and lock in that gain while working on the hard one. You seem to be doing it the other way around. If you had just said that, then at least I would have been assured that you understood the issues and I could have skipped going to so much trouble to try explaining them.

I will watch the evolution of your SR efforts with interest.

--Rik

Yes, different approach. I don't pick low hanging fruit. I look for the limiting (or most annoying) aspect, convolve it with cost/benefit analysis, and decide what to do. In this case, I considered the un-sharp 100% more limiting (AND most annoying), so wanted to tackle it first. But until actually reading and understanding the recent discussions on Super-Resolution, I did not have any tools to solve the problem.

Now I need to find a viewer that can effectively zoom in to the high res image. I like the progressive one used by the Alpa folks, recently highlighted. I like the GigaPixel one (of course, it's the epitome of cool). But these are proprietary. If anyone knows of an off-the shelf, web-capable viewer, please speak up!

And I also need to figure out my next camera move...

[Lou Jost]

Regarding an off-the-shelf viewer, you can use Google Earth to zoom in on an image. Most people I know have Google Earth already installed. Just make your image into a KMZ image overlay in an ocean somewhere. That's easy in Google Earth. Set the drawing order to some high number. Then the recipient can open it in Google Earth and zoom in and out. I've done it with 100Mb images; I am not sure what the limit is.

[ray_parkhurst]

Regarding an off-the-shelf viewer, you can use Google Earth to zoom in on an image. Most people I know have Google Earth already installed. Just make your image into a KMZ image overlay in an ocean somewhere. That's easy in Google Earth. Set the drawing order to some high number. Then the recipient can open it in Google Earth and zoom in and out. I've done it with 100Mb images; I am not sure what the limit is.

Hmm, interesting idea. What's the process for making the KMZ image overlay?