This bug is reproducible in about 5% of resets. To reproduce it, you need to reset encoder - if you are lucky enough, the bug will show up (~5% reproducible). The corrupted stream will continue to show up for each encoded frame (with those vertical lines) until you reset the encoder again. In most cases after reset it starts working fine. Then you can reset it again several times, and this bug can show up again.
Please help me to understand where might be the reason for it and how to fix it.
An example of corrupt frame is here: https://plus.google.com/photos/100915540970866628562/albums/5755521382198971601?authkey=CPau-p_7kbHynQE
Can you provide functional simulation demonstrating the issue? If it occurs only on hardware then it may well be wrong handling of external reset, e.g. not all flops exit reset at the same time if reset signal is not synchronized to clock domain used.
Thanks for response. I cannot reproduce it in the simulator, It happens only in the hardware.
I'll double check reset handling.