Reading on the web, I have found an alternative way to obtain DIC images with double contrast and background homogenization by recording two DIC images of the specimen with reversed bias. After that post processing through some algorithm will produce a high contrast image, showing features which were hiding in both the original DIC images.
When I first read of this and saw some pictures, I thought it was some high-end lab stuff, but with further readings and some tries I could obtain good results. I’m still working on this technique and improvement is still possible.
Long story short, all you need to do is to shot two images of the same sample with opposite bias which will produce opposite shadowings. No matter how you obtain it. You can slide the objective prisms, use a half lambda plate by rotating it 90°, or you can buy some more *expensive* ferroelectric liquid crystals modulator. This last choice could allow the reversal of bias with a high frequency rate. The labs could synchronize the camera frame rate with the one of the modulator, making it possible to obtain real time processed images in PM-DIC. For my home use, I think a static setup for immobile samples will do.
So the two acquired images are processed through some algorithms in ImageJ to obtain the final image.
I’m posting here a few images because I didn’t have much time to test this, but the results are already very good to me and I think it will be possible to improve by taking another image of the background with no sample, in a “clean” portion of the slide, and then subtracting it to each of the sample images to obtain a more even and clean background in both images before the processing.
This is a Sphagnum moss leaf showing the small photosinthetic cells and the large hyaline cells.
The “tail” of a Copepod. The background wasn’t clean, but the effect is still evident.
The best result was on this Diatom. The “holes” perfectly show as they really are, and not just as “dots”. The dust on the eyepiece was eliminated by background subtraction.
I guess this would work on oblique light pictures too as long as the specimen is still.
I think the quality of the final image is connected to the quality of shadowings. The closer to perfectly opposite they are, the better will be the result. I’ve just slided the prisms to obtain opposite bias, so I guess doing it in some more precise way, being sure to actually have two images with opposite shadowing would result in a better processed image.
I’m not sure this can be done with color images. I haven’t tried yet. It seems the algorithms on ImageJ only work with b/w pictures!