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

CrispyBee
Posts: 989
Joined: Mon Apr 03, 2023 11:17 am

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

Post by CrispyBee »

emilien wrote:
Thu Nov 23, 2023 3:05 pm

In this case, at 90x we have a pixel size of about 1.3 um and a horizontal FOV of approximately 3 mm for 2500x2050 pixels. Keep in mind that for the Girl with a Pearl Earring, we used a crop: we only captured the central part of the image, using 2000x2000 pixels for each tile. Also I used a 20% overlap in X and 20% overlap in Y. I will let you do the math!
In any case, I’m not sure how you end up with your calculation of 4x but please don’t compare our fully motorised XYZ high end system with a triple objective telecentric lens to a plastic USB microscope 😉
Using a 15" screen as a reference for magnification is understandable from the company history but does not relate to just about any other defined magnification typically used. Magnification is generally the purely optical magnification of a projected image - in this case how large the projection of that FOV of 3mm would be on your sensor.

2500 pixels for a horizontal FOV of 3mm is actually really not too far away from a 4x magnification for a conventional microscope lens on a regular camera sensor - for example on a camera with a 3,91 micron pixel sensor (like the Sony A6*00 cameras) the sensor is 23.5mm "long" with a resolution of 6000 pixels. So every mm has a resolution of approx 255 pixels, 3 mm have approx 766 pixels - and that's at a life-size reproduction (1x).

In order to get a FOV of 3mm projected onto 2500 pixels you need to magnify that area around 3,26x - with an additional pretty hefty overlap of 1750 pixels left and 1750 pixels right (=29% overlap on each side) without having to crop the image.
Though it would make more sense to either use a higher magnification with a higher effective resolution - or keep the magnification and effective resolution the same but use a larger FOV with a smaller overlap - which would result in fewer images necessary to cover the whole painting.

Having said that it would probably require a larger setup and it would be much more difficult to move the camera with all that additional weight etc. But that could be achieved with something like a Mitutoyo M PLAN APO HR 5X 0.21 or maybe even a Qioptiq Mag.x LD PLAN APO 5X 0.20 with a good tube lens.

There are a lot of ways to pet a cat.

rjlittlefield
Site Admin
Posts: 24209
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 »

emilien wrote:
Thu Nov 23, 2023 3:05 pm
if there are any further questions, I’m happy to answer them!
Many issues about resolution and DOF can be answered by knowing the working NA of the objective, regardless of how the system magnification is specified.

I have looked for, but never found, any info about the NA ranges of Hirox objectives. For example it seems to be absent from https://www.hirox-usa.com/products/3d-d ... pe/lenses/ .

Can you point us to documentation about NA of the various optics?

--Rik

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 »

Dear Ray, dear Rik, dear all,

I work in the field of digital optical microscopy for 20 years now and I have never heard anyone comparing our 90x to a 1x or even 4x. I’m sure you all have long experience and are all professionals in macro photography, here it was made by a digital microscope, may be the standard is just different? Out of curiosity, what is the maximum magnification in microscopy according to your calculations?

FYI, I sold to the most prestigious universities, research centers, leading companies in the field of aerospace and semiconductor, micro mechanical, biology, museum and so on in over 20 countries: they all had very good labs with optical microscopes already. Yet they decided to buy our system because it gave them better results than what they currently had. Not one of them mentioned that our magnification was wrong by a ratio of 30 or 90.

So either you are using an outdated way to calculate magnification, or you only speak about the objective and not the total magnification, or the current standard to measure magnification simply evolved.

Here are other companies producing digital microscopes: Leica DVM6 goes up to 2350x, Zeiss Smart Zoom 5 goes up to over 1000x, Keyence VHX series goes up to 6000x,… and they all use our FOV / magnification ratio standard.

Regarding 100 EUR USB microscope, they don’t use the same standard, they are cheap chinese made webcams but to be honest, they actually do a decent job for the value. It is irritating though that they are also called “digital microscopes”.

To be very honest I really don’t care about magnification and I don’t believe we are misleading, we are very straight forward with our field of view per magnification ratio. And again: no one is buying our microscopes because they “zoom more” than other technologies. We always do demonstration with customer samples, they can see exactly what lens, what zoom and what result they would get, and we sell several hundreds of systems worldwide every year.

Regarding Na, this is indeed a very good way to compare optical microscopes but I don’t have the value for this specific lens.
The reason I stated it was made with 90x magnification (not 90x objective), is as a comparison to the previous 35x magnification used on my 2018 scan.

By the way, here is an article in PetaPixels about the scan, may be you will find useful information and pictures as well:
https://petapixel.com/2023/11/21/108-gi ... gest-ever/

Ray, to answer your question about 3D of the full painting, of course, we stitched every 41106 focus stacked images not only every in 2D but also in 3D to get the entire painting in 3D. You can see some results in the YouTube video.

See one picture below showing a smaller resolution of the full painting in 3D
https://1drv.ms/i/s!Ag3uMpjM__QqhuU-78mEKLnOe1vqnQ
I tried to attach some pictures of the depth map in greyscale in attachment but I can’t attach anything higher than 1024x1024 pixels 😂
Have a nice weekend everyone!

PS:
For ray, I saw you make pictures of coins, have a look at a small scan I made a few years ago of a coin that has been turned into jewelry (cut to remove some parts of the coin by a Lithuanian girl I met, I believe the coin is from Argentina):
https://www.hirox-europe.com/gigapixel/KARMA/

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 »

emilien wrote:
Sat Nov 25, 2023 6:00 am
...
So either you are using an outdated way to calculate magnification, or you only speak about the objective and not the total magnification, or the current standard to measure magnification simply evolved.

Here are other companies producing digital microscopes: Leica DVM6 goes up to 2350x, Zeiss Smart Zoom 5 goes up to over 1000x, Keyence VHX series goes up to 6000x,… and they all use our FOV / magnification ratio standard.
...
Ray, to answer your question about 3D of the full painting, of course, we stitched every 41106 focus stacked images not only every in 2D but also in 3D to get the entire painting in 3D. You can see some results in the YouTube video.
For ray, I saw you make pictures of coins, have a look at a small scan I made a few years ago of a coin that has been turned into jewelry (cut to remove some parts of the coin by a Lithuanian girl I met, I believe the coin is from Argentina):
https://www.hirox-europe.com/gigapixel/KARMA/
Hi Emilien,

That Karma coin appears to be quite fragile the way it was cut out. I did similar things to US Mercury Dimes long ago to make jewelry. It turns out very nice with just the main feature showing.

Regarding magnification, you are correct when you say we are talking about the objective, not the total. The total magnification includes "digital magnification" and is somewhat arbitrary since it depends on the screen size...bigger screen, higher magnification. It also depends on the sensor size...smaller sensor, higher magnification. Certainly I understand why this is done, since a "90x" or "1000x" microscope sounds much better from marketing perspective than a "1x" or "4x", but it is scientifically very imprecise.

Ray

Scarodactyl
Posts: 1768
Joined: Sat Apr 14, 2018 10:26 am

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

Post by Scarodactyl »

There's a certain irony here, since the inflated magnification is entirely the wrong approach for marketing to us in particular--the marketing should instead be emphasizing how this system has an astonishingly low (effective, post-stitching) magnification for its numerical aperture. Girl with pearl earring is like 44cm long, on my aps-c sensor it's equivalent to an astonishing 0.05x/0.1 objective! (Just guessing on the NA, the only one I can have a good guess of is the OL35x which is (probably) a mitutoyo 5x/0.14).
Last edited by Scarodactyl on Sat Nov 25, 2023 12:00 pm, edited 1 time in total.

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 »

Scarodactyl wrote:
Sat Nov 25, 2023 10:43 am
There's a certain irony here, since the inflated magnification is entirely the wrong approach for marketing to us in particular--the marketing should instead be emphasizing how this system has an astonishingly low (effective, post-stitching) magnification for its numerical aperture. Girl with pearl earring is like 44cm long, on my aps-c sensor it's equivalent to an astonishing 0.05x/0.1 objective! (Just guessing on the NA, the only one I can be sure of is the OL35x which is just a mitutoyo 5x/0.14).
The numbers are a bit confusing, but I think the FOV on subject is 2mmx2mm, and Emilien confirmed this was for a 2000x2000 pixels central capture, 1.3um sensor pitch, or 2.6mmx2.6mm FOV on image, giving a magnification of 1.3x. I don't think this was done with a 5x objective, unless the numbers provided are incorrect. I initially assumed a much larger sensor, and indeed a 2.6mm sensor seems extremely small, so there may be some error there. Emilien also states the objective is telecentric, which the Mitty is not, unless some additional optics were added to make it so.

Scarodactyl
Posts: 1768
Joined: Sat Apr 14, 2018 10:26 am

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

Post by Scarodactyl »

The 5x objective is from their higher mag lens system (hr-10ce), the one used for the painting is a lower mag all-in-one (hr-1020e). I'm just blind guessing at the NA which will be lower than 0.14.

CrispyBee
Posts: 989
Joined: Mon Apr 03, 2023 11:17 am

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

Post by CrispyBee »

emilien wrote:
Sat Nov 25, 2023 6:00 am


Here are other companies producing digital microscopes: Leica DVM6 goes up to 2350x, Zeiss Smart Zoom 5 goes up to over 1000x, Keyence VHX series goes up to 6000x,… and they all use our FOV / magnification ratio standard.
https://www.leica-microsystems.com/prod ... ification/

https://cdn.standards.iteh.ai/samples/6 ... 1-2016.pdf

Yes, these are not optical or effective magnifications and should be defined as "display magnification" or "visual display magnification", otherwise it causes confusion and could be mistaken for optical magnification - which would of course be nonsense as a magnification of 2350x is not possible with optical equipment. That's well into electron microscopy, optical microscopes are usable at 1000x max.

If we were to use display magnification" or "visual display magnification" many of us would be working well beyond "3000x".

That's why digital microscopes have a bit of a questionable reputation, as they tend to distort the reality and difficulty of actual optical magnification.

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 »

I’m clearly wasting my time trying to go through each mistakes in the various comments I read here, I wish you all a lot of success and fun in your life!

Sym P. le
Posts: 347
Joined: Thu Jul 04, 2019 9:53 pm
Location: BC

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

Post by Sym P. le »

whoa! As a consumer that doesn't know his place, I clearly shouldn't inquire about optical magnification. The car has four tires, what's your problem?

Rik, feel free to delete this

CrispyBee
Posts: 989
Joined: Mon Apr 03, 2023 11:17 am

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

Post by CrispyBee »

emilien wrote:
Sat Nov 25, 2023 5:11 pm
I’m clearly wasting my time trying to go through each mistakes in the various comments I read here, I wish you all a lot of success and fun in your life!
I think you're going about this the wrong way.
This could be an excellent way to get some potential customer feedback, maybe get a fresh perspective on stuff, inquire about other methods and compare their results with yours and think about potential improvements - or if you feel you method is better, give others some feedback on how to improve their workflow.
No need to share trade secrets of course, but this could have been a valuable and interesting exchange.

Nobody says your product is bad or the results is bad - on the contrary, it's a beautiful result and a great achievement. But as we're all here interested in macro and micro photography/imaging it's a good idea to "speak the same language" and use the same terms to facilitate a mutual understanding - which is not at all degrading, only avoiding some potential misunderstandings. (That's something I had to learn myself the hard way).

For "us" the optical magnification is the basis for our work. It's something we all can relate to as it provides an equal, common ground. We all know what 1x 'means' and how it relates to getting a result. We understand the difficulties and technical challenges - and if someone says 2x or 20x, 50x or 100x we can still relate and have a mutual understanding of what was required and which challenged stood in the way of getting an image. Whether it's vibration, working distance, aperture, aberrations, distortions, diffraction, step length,..
And if we stumble upon something that doesn't line up with the experiences we've had so far, it's understandable that we start to ask questions and try to break it down to "our level".

Whether you used an optical magnification of 1x or 0,9x or 2x or 20x is interesting to us - and it doesn't devalue or degrade your work. It's just necessary in order for us to understand what you did and how you did it because we're not used to your methods and terminology.

So no hard feelings.

Scarodactyl
Posts: 1768
Joined: Sat Apr 14, 2018 10:26 am

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

Post by Scarodactyl »

emilien wrote:
Sat Nov 25, 2023 5:11 pm
I’m clearly wasting my time trying to go through each mistakes in the various comments I read here, I wish you all a lot of success and fun in your life!
I feel like we gave it a good try anyway. To put it in salesman terms maybe think a bit more about tailoring your pitch to your audience in the future. But regardless the product is cool and I'm sure your existing customers are happy.

That all said I do want to restate that Hirox could not have invented video microscopy in 1985. The below article is from a 1979 German laser journal, Laser und Optiks
Image
"the newly developed MonoZoom 4 can be used with a TV camera connection. The focus adjustment and sample control are carried out on the monitor. This reduces fatigue and makes the workplace more humane."

I do want one of the hirox systems. Not even sure what I'd do with it but the stacking and stitching still really impresses me.
Last edited by Scarodactyl on Thu Dec 07, 2023 8:39 am, edited 2 times in total.

Alex G
Posts: 22
Joined: Sun Jun 13, 2021 6:42 pm
Location: Toronto. Ontario, Canada

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

Post by Alex G »

emilien wrote:
Sat Nov 25, 2023 5:11 pm
I’m clearly wasting my time...
emilien - May I make a suggestion? I hope I speak for everyone here, but I think we all agree your rendering of Girl With A Pearl Earring is stunning. What better opportunity to make your tech achievement relatable to everyone than to handle a barrage of questions from an audience of folks who are passionate about the technology that you've mastered. If you read through these forums you'll note that conversations often travel deep into the science of the many variables that make a final image. If we challenge you it's because we're really really interested!

Your tech is amazing, so please help us understand. If the questions are odd, helping us with the answers will prepare you to answer them when you are presenting this tech to clients.

Sym P. le
Posts: 347
Joined: Thu Jul 04, 2019 9:53 pm
Location: BC

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

Post by Sym P. le »

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.

The language used by the rep is indeed standard language but not at our preferred granularity. As much as he would like us to chill and communicate only in his comfort zone, perhaps he could indulge us with some underlying facts that speak more directly to the resolving capacity of the system, which is also covered by the standard he wishes to use.

Reading through the PetaPixel article referenced by the rep,
The previous scan was made at 35x, which gave a very nice image but the depth of field at this magnification was too large to collect accurate depth information for a painting that is relatively flat.
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.

Alex G
Posts: 22
Joined: Sun Jun 13, 2021 6:42 pm
Location: Toronto. Ontario, Canada

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

Post by Alex G »

emilien, I hope you are still with us!

Suppose I have a 105mm macro lens. The lens states 1:1 (max 1x magnification).
- And suppose I have a camera with a 35mm sized sensor with 24 megapixels (6048 x 4024 pixels)
- And suppose I have another 35mm format camera with a 45 megapixels sensor (8256 x 5504 pixels)
If I use this lens at max magnification (1x) to create an image using each of these cameras, we do not need to know any external factors to know that both images had a magnification of 1x.

It's only when we add an external frame of reference that things become confusing. For me peraonally I'd like to print these at 300 DPI So:
- for the 24 megapixel image I could make a print that is 20" x 13.4"
- this is a 14x enlargement of the original
- for the 45 megapixel image I could make a print that is 27.5" x 18.3"
- this is a 19x enlargement of the original

BUT... knowing the size, resolution, and format (print, computer screen...) at which I want to view the image does not inform us about the magnification of the system that made the images. Knowing the size of the sensor and the size of the subject projected onto the sensor are all you need to know the lens magnification.

Factors like NA, FOV, resolution, number of images tiled/stacked, are all important to the quality of your completed image and your results are testament to the fact that the tech you employed achieved it's goal to an incredible degree.

I read through your FAQ and saw 'The question to ask is “What is the ratio of the size of the real object to its size in the digital image?”' May I ask even more specifically:
- In the new Girl With a Pearl Earring project, what is the size in mm x mm of the sensor used? And what is the area of the painting in mm x mm that is projected on the sensor?
With that info I think you'll answer the question "what is the lens magnification?"

After that, knowing about the subject size, total resolution including all tiles and our preferred external frame of reference we can say what the total system magnification is.

Again, thanks for embarking on this incredible project. Your work is stunning!

Post Reply Previous topicNext topic