What to do when a 10 billion Pixel scan is too small?

Here are links to articles for your reading pleasure. You may also submit brief reviews or discuss the contents of the articles.

Moderators: Chris S., Pau, Beatsy, rjlittlefield, ChrisR

emilien
Posts: 10
Joined: Tue Dec 11, 2012 10:37 am
Location: www.youtube.com/hiroxEU
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by emilien »

Hi again,

Sorry for losing my temper, a lot going on and not much time. I didn’t have the chance to read everything but I will in the coming days and come back to you with more details 😊 good night or good day depending on where you are on the planet 🌎

CrispyBee
Posts: 999
Joined: Mon Apr 03, 2023 11:17 am

Re: What to do when a 10 billion Pixel scan is too small?

Post by CrispyBee »

Sym P. le wrote:
Sun Nov 26, 2023 1:54 pm
From this I gather they moved to a higher N.A. objective to gain greater vertical resolution through a narrower depth of field and more steps. Using Visual display magnification as a metric only addresses this indirectly and only promotes the bigger equals better sales pitch. My GIMP software offers virtually infinite magnification but not a stitch of additional resolution.
The sensor is very small but it's also very densely populated, with a very small pixel size/pitch (1,3microns) and the optics - though probably very good, maybe even excellent for a digital microscope - are possibly only really good at the very center of the image with quite a few problems towards the edges; at least that's what it looks like in many areas of the final image.
You can still see the remnants of the optical defects in some parts of the picture, especially in high contrast areas such as very strong chromatic aberration - but also a definite softening/smearing in some other areas of the image:

Image

This edge and corner falloff can also be observed in other samples on their website like the HDR Auto-Optimisation preview or various other previews on their product pages, alongside noticeable distortions and vignetting - though that could only be the case for the specific lenses they used for these images and doesn't have to affect other lenses - or the specific lens used for the Vermeer painting. But it certainly creates the impression that's what most of the lenses look like before you "zoom in" and crop away the edges.
So I would guess it's a lens that has a very high resolution and correction in the center portion of the sensor (the part that's mostly used when enlarging/"magnifying" the image on a display) but a pretty strong loss in quality towards the edges. This could probably be resolved by using a smaller aperture but then the center resolution would probably not be sufficient for the 1,3 micron sensor and the max. display magnification wouldn't look as sharp and detailed.

Of course that's perfectly fine for inspection and a digital microscope and obviously it's great for generating 3D images - but it's not that ideal for a large area art reproduction and that also explains the large amount of images needed for the whole painting, as it would require a large overlap and a lot of cropping.

It just goes to show that it's really hard to create a lens that offers extremely high resolution and correction over a large sensor area all at the same time.

So I think the lens could be similar in performance and general build/design to telecentric inspection lenses for 2/3" like the the Schneider SYLVINE 0.14/1x or something even more exotic like Myutron.

Again: this is not at all demeaning or devaluing the equipment, it's actually very impressive to have a lens that can resolve that amount of detail on a 1,3micron sensor (even if it's probably mostly in the center) and then to successfully use the lens to 'scan' such a large painting is an even greater achievement!

ray_parkhurst
Posts: 3636
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by ray_parkhurst »

Sym P. le wrote:
Sun Nov 26, 2023 1:54 pm
From the leica-microsystems.com link provided by CrispyBee, I see they refer to ISO/DIS 18221 as their frame of reference. Here is a link for the 2016 standard - ISO 18221:2016.
Thank you @Sym P. le for digging that out. I have not seen that doc before, and it is very interesting that much of what we've been discussing (and more) has been standardized for some years. I plan to study it carefully. Based on the doc, I'm hoping Emilien/Hirox can provide some of the numbers for us to review. It's nice also to have a standard terminology for the discussion.

Edited to add:

In reviewing the ISO doc, it seems they want Digital Microscope manufacturers to disclose the following info to their customers:

Optical Magnification, Object to Sensor
Display Magnification, Object to Display (IMO better to state sensor and display sizes than just their ratio)
Visual Display Magnification, Referenced to 250mm distance (I don't see how the company would know this)

Optical Resolving Power Limit, in lp/mm in Object space (perhaps just disclose NA of Objective)
Sensor Resolving Power Limit, in lp/mm in Object space (seems overly proprietary. Perhaps just disclose pixel pitch)
Display Resolving Power Limit (only useful if company provides the monitor)
System Resolving Power Limit (not sure why this is needed if other numbers are given)

Useful Range of Viewing Distance / Magnification (again, only relevant if company provides the monitor)

So assuming my notes/analysis is close to correct, all the useful info that the ISO doc requests, at least as far as what folks on this forum are likely to be interested in, would be:

Optical Magnification
Sensor Size
Pixel Pitch
NA

The list is pretty much what we've been asking for, so the doc provided no big surprises. The remaining info is only useful for folks who would buy the system with a monitor so they can know what they'll be seeing.

I did find it interesting that ISO has standardized on the ratio nomenclature (ie 100:1) for optical and Display magnifications, but for the Visual magnification they standardized on the multiplier (ie 100x).

Also found interesting that the standard viewing distance is 250mm. Seems really close! I would have expected 1000mm to be standard.

enricosavazzi
Posts: 1548
Joined: Sat Nov 21, 2009 2:41 pm
Location: Västerås, Sweden
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by enricosavazzi »

ray_parkhurst wrote:
Mon Nov 27, 2023 11:31 am
[...]
Also found interesting that the standard viewing distance is 250mm. Seems really close! I would have expected 1000mm to be standard.
I cannot find a reference right now, but according to my notes the total magnification of an optical microscope (equipped with eyepiece(s) for visual observation only, not a film camera, projection screen or solid-state sensor) is based on image size, and "The image size is measured on an image projected in front of the viewers' eyes at the conventional distance of 25 cm." So the ISO document seems to be exactly in line with common microscopy practice.
--ES

enricosavazzi
Posts: 1548
Joined: Sat Nov 21, 2009 2:41 pm
Location: Västerås, Sweden
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by enricosavazzi »

Going back to the acceptability of specifying magnification as area magnification instead of linear magnification (a.k.a. lateral magnification), I can note that ISO 10934:2020(en) Microscopes - Vocabulary for light microscopy defines magnification of a microscope in two slightly different ways, depending on whether the microscope is used to produce a directly measurable image, or for visual observation of an aerial image:

3.1.90.8
lateral magnification
ratio of a given distance in the real image (3.1.75.3) normal to the optical axis (3.1.107) to the corresponding distance in the object (3.1.104)
Note 1 to entry: This ratio should be expressed in proportional form, e.g. 10:1.

3.1.90.10
visual magnification
ratio of the tangent of the viewing angle (3.1.147) of the object (3.1.104) when observed through a magnifying system with the image (3.1.75) at infinity, to that of the object when observed by the naked eye at the reference viewing distance (3.1.124) (250 mm)
Note 1 to entry: This ratio should be expressed in numerical form with the multiplication sign, e.g. ×10.


Additional definitions of magnification in more specific contexts are specified in the same ISO document, but none of them uses area magnification. PM.net is far from unique in specifying magnification as lateral magnification.

The ISO document is freely available online, e.g. at https://www.iso.org/obp/ui#iso:std:iso: ... sec:3.1.90 . For your convenience, this link leads directly to the definition of magnification.

This thread reminded me that nearly two years ago I published a page on my site ( https://www.savazzi.net/photography/poppycock3.html ), discussing several "classical" instances of misleading or fraudulent claims of microscope magnification (including the quoting of area magnification without clearly explaining that it does not refer to lateral magnification, which is a recurrent theme in microscope history), and few related fraudulent claims, including i.e. the PC board of a cheap webcam (probably equipped with a 3 Mpixel sensor) advertised by dozens of Chinese resellers as a "300 Megapixel Microscope Module Electronic Eyepiece Camera Module USB2.0 output".
--ES

rjlittlefield
Site Admin
Posts: 24220
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by rjlittlefield »

enricosavazzi wrote:
Sun Jan 19, 2025 3:05 pm
Going back to the acceptability of specifying magnification as area magnification instead of linear magnification (a.k.a. lateral magnification),
I must have missed something. I don't see anywhere in this thread that "area magnification" is discussed or even mentioned, except in the context of how many tiles were required.
PM.net is far from unique in specifying magnification as lateral magnification
I agree.. The uniqueness is not in using lateral magnification.

Instead, the uniqueness is that people at PMN tend to assume/want/require that the number should mean optical magnification between subject and sensor, as opposed to overall system magnification between subject and screen or subject and print.

Either meaning is permitted by ISO 10934:2020(en) section 3.1.90.8. That standard says simply that lateral magnification is the
ratio of a given distance in the real image (3.1.75.3) ... to the corresponding distance in the object

However, following the reference chain farther, 3.1.75.3 specifically allows the real image to be a screen. In fact that's the only example it gives:
3.1.75.3
real image
image (3.1.75) which can be received on a surface
EXAMPLE:The surface can be a screen (3.1.132).
So according to ISO 10934, it's perfectly acceptable to project a 10 mm object onto a 10 mm sensor (1X optical magnification), display the final image on a 500mm screen, and call the result "50X".

However, using "50X" in this way at PMN is likely to cause confusion at best. Sometimes that progresses to active conflict when people dig in on their own preferred usage and act as if the other usage is somehow improper.

From my standpoint -- speaking here as Admin -- forum policy is that we want clarity, not necessarily conformance to local customs. "Seek first to understand..."

--Rik

CrispyBee
Posts: 999
Joined: Mon Apr 03, 2023 11:17 am

Re: What to do when a 10 billion Pixel scan is too small?

Post by CrispyBee »

rjlittlefield wrote:
Sun Jan 19, 2025 4:47 pm
So according to ISO 10934, it's perfectly acceptable to project a 10 mm object onto a 10 mm sensor (1X optical magnification), display the final image on a 500mm screen, and call the result "50X".

However, using "50X" in this way at PMN is likely to cause confusion at best. Sometimes that progresses to active conflict when people dig in on their own preferred usage and act as if the other usage is somehow improper.

From my standpoint -- speaking here as Admin -- forum policy is that we want clarity, not necessarily conformance to local customs. "Seek first to understand..."

--Rik
Also, to quote directly from the ISO document:

3.1.90
magnification


Note 1 to entry: The type of magnification such as visual or lateral should always be specified.
Which is exactly to avoid any confusion. Otherwise the given magnification on the objective would be meaningless.

ray_parkhurst
Posts: 3636
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: What to do when a 10 billion Pixel scan is too small?

Post by ray_parkhurst »

CrispyBee wrote:
Sun Jan 19, 2025 11:10 pm

Also, to quote directly from the ISO document:

3.1.90
magnification


Note 1 to entry: The type of magnification such as visual or lateral should always be specified.
Which is exactly to avoid any confusion. Otherwise the given magnification on the objective would be meaningless.
The doc attempts to avoid confusion by specifying lateral/optical magnification be stated as a ratio (ie 1:1) while display magnification be stated as a multiplier (ie 50x). Just do that and all confusion will be gone. :wink:

CrispyBee
Posts: 999
Joined: Mon Apr 03, 2023 11:17 am

Re: What to do when a 10 billion Pixel scan is too small?

Post by CrispyBee »

ray_parkhurst wrote:
Mon Jan 20, 2025 6:55 am
CrispyBee wrote:
Sun Jan 19, 2025 11:10 pm

Also, to quote directly from the ISO document:

3.1.90
magnification


Note 1 to entry: The type of magnification such as visual or lateral should always be specified.
Which is exactly to avoid any confusion. Otherwise the given magnification on the objective would be meaningless.
The doc attempts to avoid confusion by specifying lateral/optical magnification be stated as a ratio (ie 1:1) while display magnification be stated as a multiplier (ie 50x). Just do that and all confusion will be gone. :wink:
Hm I wanted to avoid having to scratch off and repaint the text on all my microscope objectives :lol:

Post Reply Previous topicNext topic