Building a full-frame digital camera from scratch - Part 2

[@hullinstruments]

Every project you do on KZbin is always great and some of the best I've seen on the platform. Approaching problems and projects that a lot of us would love to attempt... And some of the only good documentation about these niche topics. I look forward to your projects in the future

[@CuriousScientist]

Thank you! I try to work on niche topics and try to document them carefully. There's still a lot to improve, but it is a learning process for me as well. I have some interesting projects lined up, and I hope it will be interesting for you, too. Hopefully, I will publish a few in the near future.

[@tjalocha]

To focus and to adjust integration time, move the sensor to the center of the image, put a focusing chart in the center of your composition and plot out a continuous life stream of your sensor. Focus your lens by looking for the lens position with the steepest slope in your graph. And for ideal integration time, look for a graph that uses optimally your value range without clipping. If you want a numerical reference number of how well your focus is, use max value of the first derivative of your plot as a reference. the higher the absolute number the better the focus. A good way to focus would be out of your window. Objects more than say 500 meters away are good enough to count as infinite. This is ideal too to fine tune your flange distance.

[@CuriousScientist]

Thank you for the great tips! I will consider them in the upcoming update.

[@deneguil-1618]

your videos are amazing resources! I'll use them to make this project myself once i have the time for it, I had trouble understanding everything about the data sheet since i don't have a background in electronics at all but your videos on how to drive the ccd helped a lot, tho i still need to learn how electronics work if i want to understand the reasoning behind the circuits and to make custom pcbs myself

[@CuriousScientist]

Thanks! Glad to hear that my videos are useful!

[@linusgoblin]

Once you have nailed your flange to focus nicely on infinity you will be able to just measure the distance from the lens to the subject you are capturing and dial your lens accordingly to focus headaches free. Also for what you call integration time, why not use shutter speed standard times? 1/30, 1/60 etc.. This will make it more intuitive to use with basic rules of exposition like sunny 16

[@CuriousScientist]

The integration time is based on the CCD's timing. I implemented the code for the CCD and not for the "camera". So, further optimalization will come once I make everything work properly. However, since the integration time is multiples of the minimum readout time (7.5 us or so), it might not be possible to have _exactly_ 1/60, 1/30 or so. Regarding the focusing, I have a bit different idea. I fix the flange distance perfectly, then I just bring the aperture down to f8, f11 or so and just use the hyperfocal distance between infinity and whatever the other distance becomes, based on the focal length of the lens mounted. Since I want to use the camera for landscapes, it will be good enough for me.

[@linusgoblin]

​@@CuriousScientist Yes we are talking about the same technique, for your current lens the hyperfocal values for different apertures are already written on it. Also, i have a weird idea, if you made a curved rail for the ccd you could mount the lens on an axis and connect that axis to the ccd sensor to render panoramic images. A bellow would be needed to allow the lens to move and not have light leaks. With proper timings you could even do astrophotography with it, maybe?

[@doomcrimson1680]

It is like the submarine camera from iron lung.

[@CuriousScientist]

But better! 😄

[@rammenser]

I have exactly the same idea to make a cheap Thermal Camera by using a linear IR sensor

[@CuriousScientist]

Cool! If you build your device and document it somewhere, please send me the resources. I would be curious to see it.

[@sbuntinx]

Cool project, I noticed that the pixel size of the CCD is 8x200 micron. Perhaps this could be the reason why there seems to be a blurryness, but only in the horizontal direction while in the vertical direction the image seems much sharper. Do you have a link to the original non compressed image taken with the camera?

[@CuriousScientist]

Thanks! Yeah, it could be blurry because I chose a smaller step size than the width of the pixel. And since the scanning goes along the width (200 um) of the pixels, it might cause issues. However, I have much better image that I recorded with the earlier iteration of the camera and that was quite nice. So it might also has something to do with the mechanism that moves and supports the CCD. I will soon build a new iteration because I have new components, so we will see if I can improve the image quality. Regarding the picture, I haven't uploaded it anywhere. I can try to find the raw file, but can't promise that I will find it, because it has been a while since I played with this gadget.

[@sbuntinx]

Yeah i can imagine that the motion system has be smooth to avoid artefacts. I have made a camera from a linear ccd as well but i have used a rotation mechanism instead of linear. The best results were when the rotation speed was kept constant but that requires the data transfer to be fast enough to write everything to the sd card before the next buffer is filled. Not sure if you move the stepper motor continuously while capturing light with the sensor or you start and stop the motion for every line. For my mechanism the latter option introduced to much vibration but might work different for your mechanism.

[@CuriousScientist]

I scan a line between two steps. The motor does not run while the scanning is ongoing. However, there could be vibrations for sure. In my case, the data transfer is basically immediately done, so the motor does not spend too much time with parking. The reading-stepping-reading is programmed efficiently. Once I build the new mechanism and circuit (I worked out some cool improvements), I will test it again and publish the results in a similar video as this one.

[@niklassteenackers3979]

Hey, I'm currently in a similar project but with a linear RGB CCD and more pixels to digitalise a 100 year old stereocamera. I am really learning a lot from you, thank you so much! Would it be possible to add a preview feature (so less pixels taken on X axis for a quick scan) to see if exposure and focus are right?

[@CuriousScientist]

Hi! Thank you! It is nice to hear that others are also working on and interested in similar projects. The preview feature idea is great, and it can definitely be added. I will try this in the near future.

[@hullinstruments]

Might want to link part 1 in the description or comments.

[@CuriousScientist]

Thanks, I added it. 😉

[@linusgoblin]

I love german nerds that make youtube videos.. Just the best :)

[@CuriousScientist]

Except that I am not German. 😁

[@linusgoblin]

@@CuriousScientist Ah, sorry about that!

[@CuriousScientist]

No worries, no offence taken.

[@sawerlake9820]

hi, please tell me where I can find wiring diagrams and software for this project. I want to repeat it, but there are not enough materials on the site

[@CuriousScientist]

There's no wiring diagram. I haven't published any resources because the project is still under development.

[@bakac0.]

Could you share source codes?

[@CuriousScientist]

I have a tutorial on how to drive the CCD. This project's source code is however not published.

[@halmatattaali6240]

Hello sir I like to be your student how to contact you please 🙏

[@CuriousScientist]

Hi! I don't mean to be rude, but if you can't find my contact information, you should not be my student.