I am extremely frustrated with the removal of TAGH3610 from the H7 R2 firmware, especially since that change is not documented. I need to use this family because of the number of tags it contains. My project was working fine until I upgraded the IDE, which prompted a firmware update and now I have to figure out how to get back to a working state. Where can I find how to downgrade to whatever firmware last contained TAG36H10 support?
I’m working on documentation updates right now and will update the docs to reflect it being removed.
Do you want an old firmware with the tag or a new version of the firmware with the tag? I can enable it by compiling a custom firmware for you.
Either one; I just want to run my existing script that was working fine before. Claude suggested 4.6.22 (which isn’t actually available in releases) so I tried 4.6.20 and it looks like OPENMV4 crash-loops on my H7 R2.
bootloader.zip (16.4 KB)
This firmware enables TAG36H10 and disables DATAMATRIX support so that the firmware fits in FLASH.
It’s based on v5.0.0. There are a lot of API changes in drawing methods, etc. Have Claude pull the new API from the docs that I’m in the process of updating and fixing issues. You should be able to quickly get to a working new app.
Thanks, but I’m not totally clear on how to use that zip – it seems to only have bootloader.bin and no firmware.bin. I was expecting to use Tools → Load Custom Firmware from the IDE and select some firmware.bin. Is there a different method I should be using that I can read about?
Claude needed 80k+ tokens and 15 minutes to update a 177 line script to the new API, and the firmware above runs the script at 8 fps with a memory leak. I managed to roll back the bootloader so that I could downgrade firmware and 4.4.2 runs the script at 28 fps.
You should get about 20 FPS at QQVGA in either RGB565 or GRAYSACLE on the h7 regular with a QVGA tag. I don’t know what you are doing in your script, though.
We should check if we can re-enable those tag families, given the recent fixes that freed up some flash memory. Note that we disable features on older boards because they simply run out of flash, either because a MicroPython update pulls in more features or our firmware adds/refactors a core component that requires more space. We regularly have to guess which features might be used least and disable them.
Regarding AprilTags, if anything, they should run even faster, not slower with a leak. I’d be interested in testing this if you can provide a small script to reproduce the issue.