Glare problem with 144 LED right light for microscope

[lukego]

... To what extent can I compensate for a small sensor using a reducer lens?

Could you post a image of a light object that shows the vignetting at all four corners?

Here are two pictures taken on a white sheet of paper, one at minimum zoom and one at maximum zoom (both with my usual 0.7x barlow lens.)

Your reduction lens may be losing a modest amount of light, but I am pretty sure you are losing a great deal more sensitivity (many f-stops) from the small, 1/2.8" sensor.

Is that likely to also affect depth of field? I would quite like to extend my depth of field such that I'm able to focus the full height of basic electronics components (e.g. MLCCs) instead of having to choose between seeing the top verses seeing the PCB. I'm not really clear on how depth of field fits in with other aspects like sensor sizes and reducer lenses but since we are talking about F-stops it sounds potentially related.

[jfiresto]

Follow-on question: I have a problem that the round field of view in my eyepieces is larger than the rectangle field of view in the camera. This makes me worry that the camera will crop out something important that I see with my eyes.

One way people like it, is to have the camera view out at the corners just match the limit of the eyepiece's field of the view, that is, they expand or shrink the image so that the lens+camera is just shy of vignetting (starting to shade and round the corners). That works great if you have a research microscope with expensive, plan apochromatic optics.

More modest microscopes will show optical aberrations approaching the corners, so some prefer to crop the image to the camera and have it capture the sharpest and flattest central part of the image. The photo tubes I have, apply a 1.25X crop factor to what you see in the standard 10X/21mm eyepiece. The camera only captures roughly the inner 60% of the image which is about as much as you can get sharp and flat with plain, achromatic optics.

Would a reasonable solution be to switch my 10x eyepieces for 15x? The idea being that this would increase the zoom on the eyepieces and in doing so reduce their field of view to be within what the camera sees.

Sticking to 10X eyepieces can be nice because you see more between pictures or when you are not taking them. A popular alternative is to use an eyepiece reticule that shows a box delimiting the camera's view.

[lukego]

Sticking to 10X eyepieces can be nice because you see more between pictures or when you are not taking them. A popular alternative is to use an eyepiece reticule that shows a box delimiting the camera's view.

Thank you very much for the tip about an eyepiece reticule. I naively tried to accomplish this by stringing a rectangle of dental floss across the bottom of the objective lens, but (of course) that was completely out of focus and ineffective. I see now that the solution has been lurking on Aliexpress all along but I knew not what to look for.

I am casting an admiring gaze on the 4K cameras with 4/3" Sony IMX183 sensors now. I would really appreciate the ability to take high-resolution photographs in addition to video. Do you have any words of caution about potential mistakes in shopping for such a device?

[lukego]

I am casting an admiring gaze on the 4K cameras with 4/3" Sony IMX183 sensors now. I would really appreciate the ability to take high-resolution photographs in addition to video. Do you have any words of caution about potential mistakes in shopping for such a device?

... I suppose that I can spend more time testing my Nikon D750 with the microscope to see how the image quality compares. This camera has a full-frame sensor that can be cropped to FX size. This presumably will show me the potential improvement from a microscope camera with larger sensor.

Separate issue is 1080p verses 4K resolution but that's really secondary. I only want 1080p final results but would see 4K as a nice-to-have as headroom for post-processing like rotating and cropping before concerting to 1080p.

[jfiresto]

Thank you for the images. Something is not quite right, there, especially with the first image (fully zoomed out?). Could I trouble you to repeat the experiment without the barlow lens? I like to take images of graph paper, in sharp focus at the center, as the grid also help show aberrations.

More camera friendly stereo microscopes – or their add-on photo tubes – include dual apertures that let you stop down and win more field of view. The ones I am familiar with are, I think, around f7 when wide open, and do not give much depth of field unless they are really stopped down, noticeably dimmer and starting to lose detail. I can see why stacking has become so popular.

If you think of the microscope as the lens, I tend to the notion that you want to stop down, and for that you need much more light on the sensor and a sensor that needs less light. There is a fair amount of art in getting that – which I am learning – so I am not sure what to suggest.

[lukego]

If you think of the microscope as the lens, I tend to the notion that you want to stop down, and for that you need much more light on the sensor and a sensor that needs less light. There is a fair amount of art in getting that – which I am learning – so I am not sure what to suggest.

Is there an easy way to follow your learning journey? I'm curious.

I'm also remembering that there are some "bare electronics" camera modules around that might be cheaper ways to access specific sensors. The Raspberry Pi camera for example: https://www.raspberrypi.org/products/raspberry-pi-high-quality-camera/. I have no desire for the OEM firmware in these various cameras nor their associated PC software.

[jfiresto]

Is there an easy way to follow your learning journey? I'm curious.

I follow and occasionally contribute to a couple forums

http://www.microbehunter.com/microscopy-forum/
http://www.photomacrography.net

much more to the former which seems a little silly as I neither look at microbes nor own a microscope that is good for that. Still, they are a friendly bunch and seem to tolerate my posts. Mostly I have been absorbing scattered information from the 'net and testing what works and does not.

I'm also remembering that there are some "bare electronics" camera modules around that might be cheaper ways to access specific sensors....

I have not tried the component/OEM route. Instead, I have gotten into photo enthusiast cameras that are easily adaptable to microscopes (e.g., the micro four thirds system). The ones I favor are sold in volume and, consequently, impressively cheap for the high quality sensors and image processing they provide. For example, I am currently finishing up an electronics box to control one favored by Asian camera girls. (I better use A4 button head screws on its front panel or it is going to look dowdy.)

[lukego]

I have an idea for a cheap contraption that might give stellar results: a new monocular microscope objective lens for the autofocus camera:
https://www.aliexpress.com/item/33001984896.html

This way 100% of the light would be directed towards the camera instead of being split into the eyepiece on a trinocular.

I have a cheaper camera that I could put onto the trinocular and it might actually work slightly better (has a slightly larger sensor than the autofocus.)

Maybe I could even wield the autofocus-monocular microscope hand-held? I'd point it at parts of the PCB where I want a closer look, while the PCB sits clamped under the trinocular microscope. I would need a short working distance, high adjustable magnification, and perhaps a monocular lens that's well suited to the 1/2.8" sensor if I'm able to identify one.

Crazy?

[lukego]

... if that wasn't sufficiently crazy...

How about if the monocular scope would hang from a slightly elastic cord from the main scope? And have a ~2cm working distance so that its base could be rested directly on the PCB under inspection? Then you would have one end firmly resting on the PCB at a suitable focal distance and the other end suspended from elastic so that you can maneuver it without excessive wobble.

(This still leaves the issue of image quality on the trinocular scope but this would reduce the requirements there if there's a separate solution for close inspection.)

[jfiresto]

Crazy? I do not think so. Stereo microscopes make compromises to produce stereoscopic images that a camera does not need and can not use. If someone has already paid for the microscope and its compromises are not unhelpful, it can be easier and less expensive to build on what you have got.

[lukego]

I've been informed that RasperryPi now has a $50 camera module with a Sony APS-C sensor that kicks out 1080p30. https://www.raspberrypi.org/blog/new-product-raspberry-pi-high-quality-camera-on-sale-now-at-50/. (The sensor seems to be more capable than this and the bottleneck is what a Pi can encode.) I'm further informed that a RaspberryPi Zero for $5 can make this camera module function as a standard USB webcam (http://www.davidhunt.ie/raspberry-pi-zero-with-pi-camera-as-usb-webcam/).

Sounds potentially amazing, depending on the actual performance of the sensor and video encoding?

[dreamcat4]

Yeah i have seen that vignetting before and it's the cheap CVT 1/3 reduction lens. Vignetting like this also occurs on the cheap ~$20 version 0.35x lens. However the $90 version of the 0.35x should not have that issue. From the various samples i have seen over time.

BTW the later 0.35x version is only compatible with certain microscope heads. The camera port is a different diameter somethin like 38mm. So be aware of the compatibility.

[lukego]

I've ordered the RaspberryPi HQ camera and RaspberryPi Zero. This is supposed to make a 1080p30 C-mount webcam with APS-C sensor for ~$60 (see above) and that demands investigation  . I'll see if the Pi performs better on my trinocular scope than the IMX290 autofocus camera.

I've also ordered a monocular C-mount microscope lens from AliExpress. This might be a better match for the autofocus camera to use for inspection only.

[lukego]

Just posting to share an image from some quick 0402 inspection work this morning. I'd really like to improve on this. Have to try improving focus, using a mono lens without splitter, using the RaspberryPi HQ camera once that arrives...

[lukego]

My book on photographic lighting finally arrived!!! I'm pleased as punch

Meanwhile I had an image quality win on the microscope today. I made a video of soldering a USB controller using hot air. The picture quality started gorgeous to my eye but when I brought in the hot air gun it bumped a gooseneck LED out of the way. However, I was able to quickly more-or-less match the exposure in post processing just with a single point adjustment on the base curve. Pleased!

Here's the video showing my current picture quality level. This is still with the Eakins autofocus and 1/2.8" sensor (RapsberryPi cam hasn't arrived yet.)

[lukego]

I have started experimented with a Raspberry Pi Zero W wirelessly streaming video from a HQ Camera to my PC now.

The light sensitivity is much better with the Pi camera (IMX477) than the Eakins autofocus (IMX290.) It's really night and day. My long journey trying to get more and more lumens seems to have been simply working around a deficiency in the camera that I was using.

I can't easily make an apples-to-apples comparison because there are so many tunables, however with the same shutter speed I'm able to get perfectly fine images from the Pi camera that would look like indistinct darkness on the Eakins e.g. using only the ring light with the polarizer cranked up high.

Have to do some more experimentation - I have a feeling my depth of field is shallower with the Pi and that's an issue - and have to 3D print a suitable enclosure for the Pi+camera.

[dreamcat4]

Have to do some more experimentation - I have a feeling my depth of field is shallower with the Pi and that's an issue - and have to 3D print a suitable enclosure for the Pi+camera.

Thanks for letting us know luke. I myself happen to have a zero w already doing nothing and it seemss like a good choice. Please try to get back to us later on with some photos, comparisons etc. And if you can share the 3d design files on thingverse or wherever then we would of course love to see that too.

Good job man! 

[lukego]

Here very briefly is a picture from the Pi Zero W that's taken with only my relatively weak ring light and the polarizer dialed up (which reduces brightness by about half.) This would look like a ship in the night on the Eakins (sorry I don't have an exact comparison photo handy, but I'll try to produce one later.)

[jfiresto]

I like the new camera. It is much brighter and sharper overall and even more so toward the edges.

[lukego]

Here's a video from the RPi HQ camera today. I've been staring at the screen for so long I can't tell if it's good or not anymore What do you all think?



That's without any digital enhancement. Usually I have to boot the exposure waaay up but not with this sensor. I should drop the saturation a little though I suspect.

Command used: raspivid -ag 12 -dg 1 -fps 30 -hf -vf -ss 25000 -ag 12 -dg 1 --drc high --awb sun --mode 2 -t 30000 -w 1296 -h 972 -o /dev/shm/demo.h264

(That's a special incantation to trigger all the special cases to use the full sensor binned 2x2 in the closest res to 1080p30 that I can manage. I have some hope of capturing a better image via HDMI instead of having the RPi encode H.264 but I need to wait for a micro-HDMI adapter to test that.)

EDIT: Let me add that a nice feature here is that both the field of view and exposure in the camera is much closer to what I see in the microscope eyepieces. This means I don't have to be so careful of accidentally cropping out the action on the recording. I also don't have to blind myself while trying to get enough light to satisfy the camera . Hopefully this good exposure also means that when I zoom riiight in there will be enough margin to tease out a decent image there too.

EDIT2: also note there's a slight slowdown factor on that video because it was captured at 30fps but accidentally uploaded to youtube as 25fps.

[lukego]

Today I made a video of soldering 01005 at high magnification with the RPi HQ camera on microscope trinocular port.



I'm really satisfied with this setup because it's now easy to capture a reasonable image. Just a little extra illumination, just a little slower shutter speed, just a little boost in post, no need for any extreme measures.

That video was re-encoded at lower bitrate after the RPi captured it when I cut it and so compression artifacts will be not due to the RPi. Overall it's taken quite a while to tweak the right settings on the Pi, see which ones clobber which others, and avoid glitches due to e.g. mismatched framerates through my editing pipeline. However with the right incantations it seems to work really smoothly and easily now.

For reference here is my current magic incantation to get good exposure, high encoding quality, and full sensor used in 2x2 binned mode:
raspivid --vstab -qp 18 -ag 12 -dg 1 -fps 30 -ss 22000 -ag 12 -dg 1 --awb sun --mode 2 -t 600000 -w 1296 -h 972 -o 05001.h264

I'm pretty satisfied now though probably should e.g. tune the white balance and see if a little longer/shorter shutter speed is a better brightness/blur tradeoff.

[Electro Fan]

My book on photographic lighting finally arrived!!! I'm pleased as punch

Meanwhile I had an image quality win on the microscope today. I made a video of soldering a USB controller using hot air. The picture quality started gorgeous to my eye but when I brought in the hot air gun it bumped a gooseneck LED out of the way. However, I was able to quickly more-or-less match the exposure in post processing just with a single point adjustment on the base curve. Pleased!

Here's the video showing my current picture quality level. This is still with the Eakins autofocus and 1/2.8" sensor (RapsberryPi cam hasn't arrived yet.)

I think this is one of your best so far and after following your posts I think you are on a good path.  Plz remind us of what scope and camera configuration and settings were used in this one.  Thx

[lukego]

I think this is one of your best so far and after following your posts I think you are on a good path.  Plz remind us of what scope and camera configuration and settings were used in this one.  Thx

The quality of that one is totally down to lighting. I have two LED gooseneck lights shining sideways at a sharp angle from about an inch away. The sideways lighting prevents glare and the close placement provides extra light for the camera sensor.

That video is taken with the Eakins trinocular microscope (AmScope clone) and the Eakins IMX290 autofocus camera. This camera only works well on this microscope with very strong illumination (bright to the point of absurdity I would say.)

I always use the RaspberryPi HQ camera now. The sensor performs much better in normal lighting (2x2 binned mode) and it's easy to capture a good image without being obsessive about moving the lights as close as possible. I ssh into Linux on the RPi to record directly and don't need a HDMI-USB capture device either. Much cheaper and easier to get good results, usually no post processing to boost exposure needed, highly recommended for Linux aficionados.

[dreamcat4]

hey whats the maximum resolution+framerate you can capture this way?

and have you considered to try streaming the video feed over LAN to your computer? Using something like NDI / rtmp / vlc / obs etc?

thanks 

[ChuckDarwin]

Another example of overhead ring-light vs side lighting (gooseneck leds).  Both views were instructive.