writeHD108()

template<typename CONTAINER_UIN8_T>

void fl::PixelIterator::writeHD108 ( CONTAINER_UIN8_T * out, bool managed = false )

inline

Encode pixels in HD108 format (zero allocation)

Parameters

out	Output buffer to write encoded bytes

Note

Protocol: 16-bit RGB with gamma correction and brightness control

Parameters

managed

True when the source is the colour-managed one, i.e. loadAndScaleRGB16 yields a device drive rather than a widened 8-bit pixel. Then the fixed 2.8 gamma is skipped: it would be a second shaping stage after the device solve (#4326).

Definition at line 740 of file pixel_iterator.h.

                                                                              {
        auto back_ins = fl::back_inserter(*out);
 
        // One loop for both sources. This used to be two: the managed path
        // written out by hand here, and the 8-bit path routed through
        // `encodeHD108` over `makeScaledPixelRangeRGB`. Each carried its own
        // copy of the framing -- the 8-byte start frame, the per-LED emit of
        // header + three big-endian channels, the latch -- and together they
        // made this the largest encoder body in a Blink build (481 B on
        // esp32dev, ahead of `writeWS2812`, the only one Blink runs). The only
        // thing that differed was where the 16-bit channels came from, so
        // that is now the only branch (FastLED#4402).
        //
        // Byte-identical to the iterator path by construction:
        // `ScaledPixelIteratorRGB::advance()` is exactly `has(1)`,
        // `loadAndScaleRGB` into wire order, `stepDithering`, `advanceData`,
        // once per pixel; the bytes it hands `encodeHD108` are captured before
        // the step, which is what this loop does too.
        //
        // Inline rather than an `encodeHD108_wide` in hd108.h: that header is
        // included *by* this one, before `PixelIterator` exists, and a
        // concrete `PixelIterator&` parameter is not a dependent type -- so
        // its body would be checked there and fail (#4326).
#if FL_PLATFORM_HAS_TINY_MEMORY
        // No wide source exists on TINY, so the flag selects nothing there.
        FL_UNUSED(managed);
#endif
        for (int i = 0; i < 8; i++) {
            *back_ins++ = 0x00;  // start frame
        }
        // `hd108BrightnessHeader` discards its argument and pins every gain
        // at 31, so this is 0xFF 0xFF whatever is passed (#4402).
        u8 f0, f1;
        hd108BrightnessHeader(0, &f0, &f1);
        fl::size num_leds = 0;
        while (has(1)) {
            u16 r16, g16, b16;
#if !FL_PLATFORM_HAS_TINY_MEMORY
            if (managed) {
                // The managed source has already quantized its device drive
                // once, to 16 bits. `hd108GammaCorrect` exists to widen an
                // 8-bit pixel to the 16 this chipset carries; a fixed 2.8
                // curve on top of a solved drive is the second shaping stage
                // B1 and section 6 of the spec forbid after the device
                // solve, and unlike UCS7604's `mGamma.value_or(2.8f)` this
                // one is hardcoded -- no caller could have chosen otherwise.
                loadAndScaleRGB16(&r16, &g16, &b16);
            } else
#endif
            {
                u8 b0, b1, b2;
                loadAndScaleRGB(&b0, &b1, &b2);  // wire order
                r16 = hd108GammaCorrect(b0);
                g16 = hd108GammaCorrect(b1);
                b16 = hd108GammaCorrect(b2);
            }
            *back_ins++ = f0;
            *back_ins++ = f1;
            *back_ins++ = static_cast<u8>(r16 >> 8);
            *back_ins++ = static_cast<u8>(r16 & 0xFF);
            *back_ins++ = static_cast<u8>(g16 >> 8);
            *back_ins++ = static_cast<u8>(g16 & 0xFF);
            *back_ins++ = static_cast<u8>(b16 >> 8);
            *back_ins++ = static_cast<u8>(b16 & 0xFF);
            stepDithering();
            advanceData();
            ++num_leds;
        }
        const fl::size latch = num_leds / 2 + 4;
        for (fl::size i = 0; i < latch; i++) {
            *back_ins++ = 0xFF;  // end frame
        }
    }

References advanceData(), fl::back_inserter(), fl::FL_NO_EXCEPT, FL_UNUSED, has(), fl::hd108BrightnessHeader(), fl::hd108GammaCorrect(), loadAndScaleRGB(), loadAndScaleRGB16(), and stepDithering().

Here is the call graph for this function: