I'm going to build my own large format digital camera. It will operate much like a pinhole camera.
But instead of analog film I'm going to use digital light sensors. To push down the price the sensors will be placed in a line, and then the line will be lowered step by step to take the picture in segments. Like a scanner.
And instead of a pinhole I'm going to use a lens, for shallow depth of field and more light.
The problem is resolution. To get as many megapixels as possible I need very small lightsensors and a very large projection.
The smallest sensors I've found are ~4 millimeters in diameter. If I want 1 megapixel with those the projection needs to be 4 meters wide. And that's not good.
So what I'm asking is where do I get very small sensors and very big lenses?
If you got any other solutions or ideas feel free to share.
>>994806
wow, that seems really cool project, I got something similiar in mind a few months ago, really interested, will follow.
as for ideas, I would suggest making a tiny holes for each and every sensor or try to come up with a software solution later in the process ?
>as many megapixels
Sorry, your fundamental understanding of "many megapixels" is too limited for this topic. You're better off getting a large piece of photo emulsion paper, and scanning them. By the time you get as many photosensors as you'd need, you'd be better off buying a medium format like the Pentax 645D.
>>994810
Sorry that probably came off as a little blunt. I'm saving you from heartache of a project you could invest $1,000+ in to get mediocre results. Source: I've tried something very similar in astrophotography, and ended up with a 0.05MP image with ~$500 invested by the school in CCD light sensors.
>>994806
>Like a scanner
Such cameras have been made, using scanners as line cameras.
http://golembewski.awardspace.com/cameras/index.html
>>994809
>tiny holes for each and every sensor
How do you mean?
>>994810
>>994814
I understand the difficulties. You are right, "many megapixels" is super optimistic. What I mean is I want as high resolution as possible.
The lens I have right now can produce a projection that is 3x3 decimeter. If I used the 4 millimeter sensors combined with that lens I'd get a ~0.006 Megapixel image.
I'm doing this project to build a camera, and not to take photos, really. But if this turns out to be too complicated I will probably do as you suggest, and just make it analog. There are many interesting analog photo techniques, I'd love to try out the wet plate process.
>>994815
Very helpful link, thank you. Using a flatbed scanner is probably a much more reliable option.
Although not as fun, haha.
So you want a bunch of light sensors in a line? You can use a scanner for that. Pic related is exactly what you want to build.
>>994836
If you want high resolution why not just put several cameras together so that they take a picture at the same time? That is what gigapixel cameras do. Pic related is a gigapixel camera, it's like 98 cameras stuck together.
http://disp.duke.edu/projects/mosaic/cam002.html
Making a gigapixel camera from a bunch of individual photosensors is dumb.
>>994806
As >>995058 is driving at, I beleive, you would do better to have a single sensor that steps across a horizontal line, then increments one line down and does it again. If you are putting a bunch of 4mm wide sensors together then you are wasting a lot of horizontal space, when you could have a single sensor taking 0.1mm steps in the horizontal direction and simulate 40 sensors in the same distance as one of your actual sensors.
Why not build an apparatus that takes pictures on analog film which then is being scanned into digital form immediately?
Analog film has superb image quality and "resolution", all that you need then, is a good scanning device for the analog film.
Sure it's complicated, but the whole project is kinda complicated, so there's that.
>>995064
oh fuck no, I'm saying one can make a camera by putting a lens in front of a modified scanner.
If OP really wants resolution, this is fucking dumb though. The best thing to do is get a bunch of smart phone cameras, remove lenses from them, and arrange them in a grid in front of a lens or lenses. This is what super high resolution cameras do.
Or OP can be a cheap bastard and mount an SLR on a pan-tilt rig and take a bunch of photos and stitch em' together:
https://www.raspberrypi.org/blog/gigapi-a-raspberry-pi-rig-for-gigapixel-photography/
So you've invented the digital camera
So I've decided to go with a scanner instead of lightsensors. I've got a new lens from an overhead projector that I think will suit this project nicely, and tomorrow I will get word from my school if there's a scanner available that I can borrow.
So with a bit of luck I've already got all the equipment sorted.
What's left now is to build a rig.
My vision right now is to make it similar to an old folding camera.
The rail should be no problem, but making the bellows could be a problem.
Anyone have any ideas on how to make bellows, or a substitute for bellows?
>>995804
>how to make bellows
fold some thin card then tape it together
>scanner
ideally you want the scanner to orbit your lens rather than scan linearly.
>>995059
>take a picture at the same time
That's great and all, but I'm interested in where the hell they get a lens like that, and how do they solve the communication problem of hooking up 98 cameras to a computer.
You'd saturate the bus real quick.
>>995822
Just network them, lots of microcomputers talking gigabit Ethernet.
>>994806
>To push down the price the sensors will be placed in a line, and then the line will be lowered step by step to take the picture in segments. Like a scanner.
You won't save a dime this way, also the rest of the post seems to show you have no knowledge at all about optics. So my tip for you is to learn about optics so you can see how almost all of the conclusions you made in your post are wrong, and how living up to them will not yield a camera. And maybe after that you'll find a way to make your camera obscura.
