diff --git a/Disco_Diffusion.ipynb b/Disco_Diffusion.ipynb index 7d6e205..69b29e8 100644 --- a/Disco_Diffusion.ipynb +++ b/Disco_Diffusion.ipynb @@ -1007,6 +1007,7 @@ " return (input - input.clamp(-1, 1)).pow(2).mean([1, 2, 3])\n", "\n", "stop_on_next_loop = False # Make sure GPU memory doesn't get corrupted from cancelling the run mid-way through, allow a full frame to complete\n", + "TRANSLATION_SCALE = 1.0/200.0\n", "\n", "def do_3d_step(img_filepath, frame_num, midas_model, midas_transform):\n", " if args.key_frames:\n", @@ -1025,8 +1026,7 @@ " f'rotation_3d_z: {rotation_3d_z}',\n", " )\n", "\n", - " trans_scale = 1.0/200.0\n", - " translate_xyz = [-translation_x*trans_scale, translation_y*trans_scale, -translation_z*trans_scale]\n", + " translate_xyz = [-translation_x*TRANSLATION_SCALE, translation_y*TRANSLATION_SCALE, -translation_z*TRANSLATION_SCALE]\n", " rotate_xyz_degrees = [rotation_3d_x, rotation_3d_y, rotation_3d_z]\n", " print('translation:',translate_xyz)\n", " print('rotation:',rotate_xyz_degrees)\n", @@ -1423,10 +1423,8 @@ " else:\n", " image.save(f'{batchFolder}/{filename}')\n", "\n", - " # TODO(VR): trans_scale wasn't set in the PR. This needs to be set to a reasonable value!\n", - " trans_scale = 1.0\n", " if vr_mode:\n", - " generate_eye_views(trans_scale, batchFolder, filename, frame_num, midas_model, midas_transform)\n", + " generate_eye_views(TRANSLATION_SCALE, batchFolder, filename, frame_num, midas_model, midas_transform)\n", "\n", " # if frame_num != args.max_frames-1:\n", " # display.clear_output()\n", @@ -2030,7 +2028,6 @@ "#======= VR MODE\n", "#@markdown ---\n", "#@markdown ####**VR Mode (3D anim only):**\n", - "#@markdown EXPERIMENTAL ALPHA: Need to look into trans_scale value\n", "#@markdown Enables stereo rendering of left/right eye views (supporting Turbo) which use a different (fish-eye) camera projection matrix. \n", "#@markdown Note the images you're prompting will work better if they have some inherent wide-angle aspect\n", "#@markdown The generated images will need to be combined into left/right videos. These can then be stitched into the VR180 format.\n", diff --git a/disco.py b/disco.py index 9ec4a32..2029b1a 100644 --- a/disco.py +++ b/disco.py @@ -969,6 +969,7 @@ def range_loss(input): return (input - input.clamp(-1, 1)).pow(2).mean([1, 2, 3]) stop_on_next_loop = False # Make sure GPU memory doesn't get corrupted from cancelling the run mid-way through, allow a full frame to complete +TRANSLATION_SCALE = 1.0/200.0 def do_3d_step(img_filepath, frame_num, midas_model, midas_transform): if args.key_frames: @@ -987,8 +988,7 @@ def do_3d_step(img_filepath, frame_num, midas_model, midas_transform): f'rotation_3d_z: {rotation_3d_z}', ) - trans_scale = 1.0/200.0 - translate_xyz = [-translation_x*trans_scale, translation_y*trans_scale, -translation_z*trans_scale] + translate_xyz = [-translation_x*TRANSLATION_SCALE, translation_y*TRANSLATION_SCALE, -translation_z*TRANSLATION_SCALE] rotate_xyz_degrees = [rotation_3d_x, rotation_3d_y, rotation_3d_z] print('translation:',translate_xyz) print('rotation:',rotate_xyz_degrees) @@ -1385,10 +1385,8 @@ def do_run(): else: image.save(f'{batchFolder}/{filename}') - # TODO(VR): trans_scale wasn't set in the PR. This needs to be set to a reasonable value! - trans_scale = 1.0 if vr_mode: - generate_eye_views(trans_scale, batchFolder, filename, frame_num, midas_model, midas_transform) + generate_eye_views(TRANSLATION_SCALE, batchFolder, filename, frame_num, midas_model, midas_transform) # if frame_num != args.max_frames-1: # display.clear_output() @@ -1972,7 +1970,6 @@ frames_skip_steps = '60%' #@param ['40%', '50%', '60%', '70%', '80%'] {type: 'st #======= VR MODE #@markdown --- #@markdown ####**VR Mode (3D anim only):** -#@markdown EXPERIMENTAL ALPHA: Need to look into trans_scale value #@markdown Enables stereo rendering of left/right eye views (supporting Turbo) which use a different (fish-eye) camera projection matrix. #@markdown Note the images you're prompting will work better if they have some inherent wide-angle aspect #@markdown The generated images will need to be combined into left/right videos. These can then be stitched into the VR180 format.