Is there any more progress on a driver (of any kind) for the Image Sensor Interface? Atmel Norway said that it should be available but can't find it anywhere.
Who at Atmel said that? I've heard rumors of an experimental driver having been written (mentioned in passing during a conversation regarding v4l2) though I've never seen it in a public place.
Maybe if we smile sweetly someone at Atmel can release whatever is currently written, if for no other reason than a bit of beta testing?
I've only seen what's in the thread titled
STK1000 / AVR32 Camera support ?
There was a comment (in response to your question) by user 'hce' below
"AFAICT the ISI driver ended up as a v4l2 driver, I think avr32 support can give out the driver "
Maybe I saw that 'hce' was from Trondheim in Norway and assumed 'hce' worked for Atmel.
That's the comment I'm referring to and yes, hce does work for Atmel. Have you contacted Atmel support (email@example.com)? I don't have an immediate need so I was just going to chill until the driver found its way in to an Atmel patchset before playing with it.
I now have the ISI driver (supplied by those nice guys at Atmel) working with an LCD display on my NGW100. Looks pretty good! There does seem to be a problem with the driver operating in ITU656 embedded codes mode, I think there are some problems with the ISI (like SOF IRQ flag not clearing and the codec acknowledge bit not working properly) which may be why the ISI driver hasn't been included so far.
So you have video from a CMOS sensor being streamed into the framebuffer and displayed on the LCD ?
Could we get Atmel to include a working demo of that as part of BSP 3.0 ? and maybe even with optional MPEG4 compression / decompression ? Am I dreaming?
Have an eye over http://avr32linux.org/twiki/bin/...
There is a patchset there ready for testing, albeit with some limitations. It doesn't include the preview path but you can read() and frame and write() it to /dev/fb0. You will need to write a small camera driver (to power the thing up/down/ set options etc) if you aren't using an Atmel CCD (for which there is already a driver written).
Squidgit, thanks, that appears to be a big step in the right direction. Atmel really owes you for helping so much towards their tech support!
To the nice folks at Atmel: We're generally very impressed with your products and have several AVR based designs in progress. We're also trying to evaluate AVR32 to see if it makes sense as a platform for smart security cameras. The alternative is the fixed function MPEG4 chipset, but we believe that a lot of added functionality and value can be added with a powerful processor like the AVR32.
Plus my giving support here means that if I'm in Seattle any time I'm owed some beers and if I'm in Trondheim then I've been promised a hike in the mountains and a bottle of whiskey (totally gonna take you up on that hce ;) ). I'm happy with that!
SimonFisher, your results are very interesting. Which camera are you using, and which version of the Linux kernel and ISI patches. Did you do any modifications to the atmel-isi.c driver? Can you upload your camera driver if you made one?
Here is one you can try if you havent already.
Life's to short for waiting on slow CPU's
In answer to the earlier question, the camera was the VS6524 from ST, the driver was a modified version of one of the camera drivers that came from Atmel support (just a few i2c register value changes). The ISI driver was also the one emailed from Atmel. The drivers on avr32linux.org look more developed than the ones I have (they've done pretty much the same as I did withthe codec enable bit), and I understand the VS6524 is being canned.
I hope this helps
© 2020 Microchip Technology Inc.