Huh. Old version fork, I’m betting. Or who knows!?!
Anyway: yes. If you want to pipe video, an ATmega is unlikely to make you happy, but you can do the data transforms if you wish.
Yeah, but the imgur image you posted actually makes sense to me … The stuff that Gimp generates now on the other hand … not so much.
- look at the three lines at the very bottom …
The GIMP direct one is choosing the most compact representation for each character: byte values that are printable ASCII are coming out as plain characters, e.g., “J”, and non-ASCII-printable byte values are coming out as escaped numeric values in decimal, e.g. “\22”. Yes?
No idea. 
I’d have to do some digging to figure it out.
@Ashley_M_Kirchner_No why push it out as C source? Why not just read the BMP direct off the card… the only reason I can come up with is to save time/cpu% overhead?
Because, at the time, that’s what I came up with for fast reading of 3 bytes for each RGB set, without the overhead of a 24-bit BMP file which has more details. Moving forward, I may move to that format only because it’s easier on the end-user. But I haven’t made up my mind yet.
I think I found the frame latency to pull ~50px off the card and push it out to the strip, one pixel at a time (not memcpy) was about 15msec per frame. Not from my notes, just off the top of my head.
Yeah, I’m reading and pushing 48 pixels (RGB data) out in about 500 usecs by choice.
I can push that down to about 200 usecs without any trouble (using LPD8806 drivers.) The only hiccup is that every 4th read, it slows to about 1,200 usecs because of the SD read. But, it’s hardly noticeable.
^^ spam?
Also: 500 uS? got-damn son! In my POV photographic experiments (AKA checking out my stuff to see how cool it looks.) I have definitely seen that the 2811 is not fast enough to keep up with poi. It’s acceptable for double staffs though, at least the way I do them. I’d love to see a 5050 RGB SMD with the 8806 or or similar fast-as-fsck chip in on die.
Staffs yes, they don’t spin as fast (well, they could with the right person.) But a poi can be spun rather fast and if you don’t have a fast refresh, you’ll be stretching the image.
My next goal is to detect how fast it’s spinning and apply a speed grade to the refresh rate. It’d be so much easier with a hall effect sensor and a strong magnet but alas, that doesn’t work for a poi.
As for having an 8806 integrated with a 5050 … who knows. For some reason there only seems to be WS28x ones.