Thx again, I tested the new version these days - w/o issues and the performance benefit is noticable on E.
Haven't really compared with TEST_FRAMES on the previous build, but on the new version I'm getting solid 25fps (GFX_LIB=1; CAPTURE_SIZE=2), which is more than I hoped for
I strongly recommend you to give WLED a spin, imo it's yet another brilliant example of a group of individuals investing their free time to build something absolutely genius to share with the community. Also (My personal flicker issue aside) it integrates perfectly with libSambilight - i.e. LEDs are showing WLED effects when TV is off and automatically switch to Adalight input when its received. And in both cases you have full direct control of the ESP with a lot of features via web GUI.
Rgd. Flicker:
- I agree it's extremely unlikely to be directly libSambilight-caused and might have electrical reasons
- However, its definitely not voltage on the strip... because
a) The flicker occurs regardless of position on the strip (as mentioned you can set offsets on WLED; I managed to move the flicker from end of the strip to beginning of the strip that way)
b) WLED can display a preview of the strip via GUI - and the flicker is noticable on the web preview as well.
On your remarks rgd. input timeout / current limiting by WLED: Both are configurable / deactivatable on WLED, and I deactivated both, but the issue occurrs regardless.
So, imo 2 suspicions..
Either there's an physical issue on the (analogue) connection between TV and ESP via my FTDI, either the FTDI itself or some magic like ground loop etc.
Or like you say, WLED has some bug in that area. Looking at WLED code, this might make some sense. I'm not a great programmer but comparing WLEDs serial code with your ino sketch shows it has a bit more complexity at that end as WLED integrates TPM2 & has outbound serial functionality as well which might overlap. Also, the behavior I'm seeing could make sense considering WLEDs flow (see below) -> the pixels with wrong color seem to be displaying the "complementary" color to the one they should be displaying, i.e. most of the times red is displayed where it should be green. In addition, I could not manage to use WLEDs latest build as that crashes the ESP in conjunction with Adalight in (I'm using a modified version of 0.13.0-b6 instead).
https://github.com/Aircoookie/WLED/blob ... serial.cpp
I dont have a ESP8266 to test at the moment, only ESP32s, and the one I'm using here is nicely installed behind my TV. But I'll keep digging and report back. For now I'm out to enjoy the unusual, almost "Istanbulish" weather in Germany