Fix errors in DDP implementation

[coral]

I think these values are wrong given the spec.

0x0A in binary is 0b00001010

Spec is:

Bits: C R TTT SSS
C is 0 for standard types or 1 for Customer defined
R is reserved and should be 0.
TTT is data type
000 = undefined
001 = RGB
010 = HSL
011 = RGBW
100 = grayscale
SSS is size in bits per pixel element (like just R or G or B data)
0=undefined, 1=1, 2=4, 3=8, 4=16, 5=24, 6=32

So in this case, 0x0A maps to 010 which means 2, 4 bits per pixel
but really what the constant wants to represent is 24 bits, which should be 110 so 0x0D and 32 bits should be 0x1E

[blazoncek]

If this is true then it should be fixed upstream. Please contact @forkineye.

[coral]

@blazoncek This code is not in the upstream but was added inside the WLED repo ( https://github.com/forkineye/ESPAsyncE131/blob/7ff3c945e6750308db6619814b7e75f585d60aab/ESPAsyncE131.h#L53 there is no reference to DDP in the upstream repo)

If i use git blame the DDP support was added in WLED here https://github.com/Aircoookie/WLED/blame/cd82a343928622d3a19c696e1e028b987e29c871/wled00/src/dependencies/e131/ESPAsyncE131.h#L56

[blazoncek]

In such case let's see what @Aircoookie has to say.

[blazoncek]

Looks like the specs were updated after the change to WLED was made.
I do remember @Aircoookie talking to DDP authors about extending/defining RGBW support.

[coral]

Looks like the specs were updated after the change to WLED was made. I do remember @Aircoookie talking to DDP authors about extending/defining RGBW support.

Just to state the obvious, this is the spec http://www.3waylabs.com/ddp/ I've been using.

You are right, using waybackmachine https://web.archive.org/web/20200223131548/http://www.3waylabs.com/ddp/ the spec uses a different alignment which tracks. The older spec has a different alignment but even then I'm not sure that the code is valid, the older spec with this aligns to 8 bits per pixel so maybe there was confusion if the spec referred to pixel or channel?

[Aircoookie]

Hi! I'm 99% certain that bits per pixel element (like just R or G or B data) in the spec refers to bits per channel, not bits per pixel. Still, great thing you noticed the spec update, we should update the byte to 0x0B for 8 bit per channel RGB and 0x1B for 8 bit per channel RGBW respectively.