For the second time, I agree with you on this.
I thought it was just a slight minor issue on mine but it seems that you have it too. It is a bit misleading when trying to get where we are on the LFO curve because the triangle shape never really goes dark. Which is odd because it’s the only shape (along with sample & hold) that actually goes “below” the initial modulated value. So yeah, it’s cosmetic but also a good visual feedback when playing. If there is any possibility to address that in the next update that would be great !
I agree fully... if you look at the saw LED feedback at very low frequency, it's clear that the light intensity is controlled in a low resolution state... maybe 8 or 16 discrete steps... usually LEDs are controlled in intensity with PWM, so the resolution of that PWM modulation must be very low... probably cost saving reasons which to me is fine...
But I think that the response from the LED is not linear, but rather logarithmic, and that the lowest resolution in this case always happen in the lowest values... this can be seen with the slow frequency sawtooth waveshape... it has some rather rough changes when the sawtooth reach zero... so much that it even looks erratic sometimes in behavior when the rate is slow.
Maybe this is why the Triangle only use maybe the upper half of the values... to not look weird when reaching all the way down to zero... but all is of course speculation...