Hi everyone, i printed a high resolution microscope. I am posting pictures i took from microscope. please advice how can i do it better.
1 Like
@Anurag09 it looks as though you have a very good build there. What are the details of your setup? What objective are you using, and what version of the OpenFlexure Software? What are you using for stitching?
The images look very good in many ways. The features look sharp (at least as far as we can see when the large scans are made to a file size that fits on the Forum). The focus looks good across the images, and stitching seems to work well. What aspect do you particularly want to improve?
The main thing that stands out in the stitched scans is colour banding. This is a consequence of using the Pi Camera v2. The design of that camera to work as a normal camera with a short focal length lens means that we get desaturation of colours away from the centre of the image when we use it in a microscope configuration. This is discussed in Flat-Field and Colour Correction for the Raspberry Pi Camera Module | Journal of Open Hardware. As shown in that paper, it is possible to unmix the colours and regain most of the colour, but that would need to be in post-processing. The OpenFlexure software v2.11 (which I gather that you are using) applies the lens shading and flat field correction from that paper, but not the unmixing. In principle you need to measure the mixing on your actual system, but @j.stirling has had some success just using the unmixing determined in that paper and applying it to images from other microscopes. Addressing the colour banding is something that the core team is working on for v3.0.0 of the OpenFlexure software.
Thanks for the encouragement sir. I am using labomed 40x lens. On the software part I am using
Hello sir, thanks for the appreciation.
Yes i was talking about the colour blanding. Will it improve in HQ camera. The setup i am using is OpenFlexure Server v3.0.0-alpha 4. Objective is from labomed 40x.
Being on v3.0.0-alpha4 and the following alphas will get the most recent developments, including support for the HQ camera.
As you say, the HQ camera avoids the problems with colour saturation and banding in scans. It is not quite hitting some of the other image quality metrics I think.
There is a scan from HQ camera Christmas has come early! Its a new v3 server alpha release - #2 by adamglia. To get HQ working and to understand some of the limitations, you will need to read all of the posts in that thread down to post 13.
With your existing setup, you would get improvement in colour with more overlap of images, so that you only use the centre of each image. Of course that is then slower.