Merge 76909f1dfa into 0fd10ffa09

fix: use .cpu() for waveform conversion in AudioFrame creation (#11787 )
Update readme rdna3 nightly url (#11937 )
2026-01-30 08:10:21 +08:00 · 2026-01-18 02:30:24 +01:00 · 2026-01-17 20:18:24 -05:00 · 2026-01-17 20:18:04 -05:00 · 2026-01-16 14:52:18 +01:00
3 changed files with 5 additions and 4 deletions
--- a/README.md
+++ b/README.md
@ -240,7 +240,7 @@ These have less hardware support than the builds above but they work on windows.

 RDNA 3 (RX 7000 series):

-```pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx110X-dgpu/```
+```pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx110X-all/```

 RDNA 3.5 (Strix halo/Ryzen AI Max+ 365):

--- a/comfy/k_diffusion/sampling.py
+++ b/comfy/k_diffusion/sampling.py
@ -1322,9 +1322,10 @@ def sample_dpmpp_2s_ancestral_cfg_pp(model, x, sigmas, extra_args=None, callback
            r = 1 / 2
            h = t_next - t
            s = t + r * h
-            x_2 = (sigma_fn(s) / sigma_fn(t)) * (x + (denoised - temp[0])) - (-h * r).expm1() * denoised
+            uncond_denoised = temp[0]
+            x_2 = (sigma_fn(s) / sigma_fn(t)) * (x + (denoised - uncond_denoised)) - (-h * r).expm1() * denoised
            denoised_2 = model(x_2, sigma_fn(s) * s_in, **extra_args)
-            x = (sigma_fn(t_next) / sigma_fn(t)) * (x + (denoised - temp[0])) - (-h).expm1() * denoised_2
+            x = (sigma_fn(t_next) / sigma_fn(t)) * (x + (denoised - uncond_denoised)) - (-h).expm1() * denoised_2
        # Noise addition
        if sigmas[i + 1] > 0:
            x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * s_noise * sigma_up
--- a/comfy_api/latest/_input_impl/video_types.py
+++ b/comfy_api/latest/_input_impl/video_types.py
@ -374,7 +374,7 @@ class VideoFromComponents(VideoInput):
            if audio_stream and self.__components.audio:
                waveform = self.__components.audio['waveform']
                waveform = waveform[:, :, :math.ceil((audio_sample_rate / frame_rate) * self.__components.images.shape[0])]
-                frame = av.AudioFrame.from_ndarray(waveform.movedim(2, 1).reshape(1, -1).float().numpy(), format='flt', layout='mono' if waveform.shape[1] == 1 else 'stereo')
+                frame = av.AudioFrame.from_ndarray(waveform.movedim(2, 1).reshape(1, -1).float().cpu().numpy(), format='flt', layout='mono' if waveform.shape[1] == 1 else 'stereo')
                frame.sample_rate = audio_sample_rate
                frame.pts = 0
                output.mux(audio_stream.encode(frame))
Author	SHA1	Message	Date
amatiramisu	767ced13d0	Merge `76909f1dfa` into `0fd10ffa09`	2026-01-18 02:30:24 +01:00
Theephop	0fd10ffa09	fix: use .cpu() for waveform conversion in AudioFrame creation (#11787 ) Some checks are pending Python Linting / Run Ruff (push) Waiting to run Details Python Linting / Run Pylint (push) Waiting to run Details Full Comfy CI Workflow Runs / test-stable (12.1, , linux, 3.10, [self-hosted Linux], stable) (push) Waiting to run Details Full Comfy CI Workflow Runs / test-stable (12.1, , linux, 3.11, [self-hosted Linux], stable) (push) Waiting to run Details Full Comfy CI Workflow Runs / test-stable (12.1, , linux, 3.12, [self-hosted Linux], stable) (push) Waiting to run Details Full Comfy CI Workflow Runs / test-unix-nightly (12.1, , linux, 3.11, [self-hosted Linux], nightly) (push) Waiting to run Details Execution Tests / test (macos-latest) (push) Waiting to run Details Execution Tests / test (ubuntu-latest) (push) Waiting to run Details Execution Tests / test (windows-latest) (push) Waiting to run Details Test server launches without errors / test (push) Waiting to run Details Unit Tests / test (macos-latest) (push) Waiting to run Details Unit Tests / test (ubuntu-latest) (push) Waiting to run Details Unit Tests / test (windows-2022) (push) Waiting to run Details	2026-01-17 20:18:24 -05:00
Alex Butler	00c775950a	Update readme rdna3 nightly url (#11937 )	2026-01-17 20:18:04 -05:00
amatiramisu	76909f1dfa	Fix dpmpp_2s_ancestral_cfg_pp: save temp[0] before second model call temp[0] (uncond_denoised) was getting overwritten by the second model call before being used in the final x calculation.	2026-01-16 14:52:18 +01:00