Cool project. Just as an idea, you could implement the keys of the most common matrix configurations preloaded and detect them, so the setup is easier, it could be like setting up a keyboard in most linux distros. Something like: Press 1. Does you keypad contain #? press it or wait 5 seconds to skip. Press 8. Does you keypad contain B? press it or wait 5 seconds to skip. Set up complete
I thought about that and I'm totally game -- but based on my survey of keypads I'm not sure there *are* most common configs! The sample ones I bought for testing were often reeeeeal random. But if you've got suggestions on common layouts let me know!
I suppose this is Kinect v2 (for Xbox One)? I'm having a problem connecting mine to my Macbook Pro 13" retina 2015. I have the microsoft's adapter and the macbook has USB3 ports, but I can't get any power through to the Kinect. Any help?
Impressive! It makes me remembering when I was trying to generate composite video with an PIC16F877... (8 bit MCUs are losing space for these new inexpensive 32bit micros).
Impressive! Nice to see what these "modern" MCU's are capable to do by bit-banging VGA :) You should check Craft by lft if you haven't seen already. It's my all time favourite MCU demo with 20MHz, 8-bit, 1KB RAM and 8KB ROM and fully programmed with ASM. Just insane.
Bob Lynas That's an STM32F407. It'd work on the 41x/42x/43x parts too, though the 42x/43x have video hardware so you could save yourself some time by just using that. ;-)
Did you ever share the code anywhere? Or an approach? I've done VGa a few times in various ways on the avr8's, and done it brute force and slightly lesds brute force on the stm32 but not too well there.. the ARM chips doing a lot of caching and such means some jitter. Looking at doing crazy timers and feeding the DMA just-so should get a more tight VGA, but srtill working on that. But your little demo eating up only 100MHz? Good lord, thats pretty efficient :) Curious if youre' using any of the stm32 dirty tricks, or just timers and DMA? (on avr8, using the SPI to spit out the channels worked pretty well for mono, and colour worked well just due to the solid opcode timings.) Using ethernet or fsmc or other devices to spit out VGA might be a trick, I dunno... Just curious on your approach :)
Jeff Mitchell Sure did -- github.com/cbiffle/m4vgalib-demos/ I'm also working on a series of blog posts describing the internals in detail, in celebration of the release of my later graphics demo, Glitch: cliffle.com/article/2015/06/05/introducing-glitch/
Nice, can't wait to read more. A man after my own heart :) I ended up building a goofy (but fun!) little computer (see www.zikzak.ca) .. Z80 cpu with stm32 as gpu. Bitbanging the memory bus (didnt want to 'cheat', and 100pin is still hand solderable) really holds me back. Bitbanging the vga out is also limiting.. But I've still not gotten around to trying via fsmv or some peripheral, which shoukd get better. I'm doing 330x240 at colour.. But 800x600? Damn, thats nice. (Especially given the limited ram for framebuffers) I got sidetracked building a 3d printer so I could make a case for zikzak snd its carts.. Finally got that nearly sorted out (need to make an enclosure for the printer to keep heat control better) and then finally back to zikzak hacking (need to write an OS...) ... Good times. Cocktail parties too, though :) Looking forward to your tricks.. I shall have to check out your library!
Jeff Mitchell Very cool! I've been wanting to build a little computer with this code as its GPU for a while, but I kept getting hung up on how to efficiently move data in and out of the STM32. This is part of why I've done so much work on DMA with GPIO -- if I can just get it to act as a parallel bus slave.... which it sounds like you have! Awesome.
Great work ! Did you use SDIO for the output generation ? And DMA clocked by timer which is triggered by horizontal clock timer ? Or DMA clocked from SDIO ? I was trying it by GPIO and DMA and it's slow max resolution I have is 400x600 (half of 800x600) with much higher clock frequency.
Wow this is amazing. I'm a traditional artist moving back into computer visualization creativity. I've been tinkering with computer art and animation for around 20 years. I've recently purchased the Kinect sensor for windows and working with OpenNI. I am interested to know if you have done anything regarding using a projector. Could a projector be used to project the red color on the dancer in real time as she dances?
oooo. it's nice to see practical non-theoretical uses for this stuff. i'm sick of seeing stuff like printed shoes that on-one would actually choose to wear.
Is it possible to do the same thing with a normal digital camera? I'm aware that the Kinect-device is different from normal cameras, but if I DL the software, would I then be able to make the same thing you did, with a normal D-camera?
This is the kind of 3D I like to see, I can see a new form of motion photography that uses two camears that are much like kinect (one for taking the video image, and another to simulate a 3D effect where the camera scans where the person's head is so when they look around the screen it has a looking through a window 3D effect) I now have a dream for future art galleries using kinect.
@nouchetabar - an excellent question! My current code assumes a static camera location and gathers data over time to compensate for noise. I was planning on tilting the camera to gather more depth data next. The general technique you're describing, where the camera moves, is called SLAM, for Simultaneous Localization and Mapping. It's my long-term plan here, but it's quite a bit more complex!
Thanks for the kind words, everyone! The software has come quite a ways since I posted this video -- I'll try to get a cooler video up shortly. For a preview of some advanced features check out my "Kinect vs. Mirror" video.
