AY-6975 Consolidate retimes computation.

[robin-ynput]

Changelog Description

help resolve ynput/ayon-hiero#29
help resolve ynput/ayon-flame#28

• FreezeFrame does not work due to wong available media range boudary computation
• Negative speed is not always accurate
• TimeWrap:
  • default setup seems to go through OK
  • when length attribute is used/animated publish does got go through because the retime operation is expressed as otio.AnyVector which is none-serializable by integrate asset plugin
  • when input frame is not animated (freeze frame if I'm correct), it fails with an index error during otio subset collect.
• LinearTimeWarp produce correct output clip (tried speed = 0.5 and speed = 2), retime information can be properly gathered from Nuke.
• Retiming is not properly applied on reviewable.

Additional info

This needs to be tested alongside:

• More consolidations on Hiero OTIO exporter ayon-hiero#40 (hiero)
• More consolidations on Flame OTIO exporter ayon-flame#47 (flame)

Testing notes:

1. Run the automated tests
   .\tools\manage.ps1 run pytest .\tests\

[ynbot]

Task linked: AY-6975 Broken retiming calculation

[jakubjezek001]

Tested current dev and all is passing fine

===================================== 28 passed, 17 warnings in 1.11s =====================================

[iLLiCiTiT]

Based on the code changes, can happen that conformed_source_range.duration.value * abs(time_scalar) is less than 0? If not then this would be preferred:

        offset_duration = conformed_source_range.duration.value * abs(time_scalar)
        # Offset duration by 1 for media out frame
        # - only if duration is not single frame (start frame != end frame)
        if offset_duration > 0: 
            offset_duration -= 1

From code reading perspecive more clear what is being handled.

NOTE: comment duration 1 frame -> freeze frame -> end = start + 0 was confusing for me, but feel free to keep it if it makes more sense.

[iLLiCiTiT]

Reading if not 0 <= new_idx < len(frame_range): is pretty challenging...

Suggested change

	frame_range[idx] = round(frame_number + tw["lookup"][idx] * time_scalar)
	# Consecutive timewarp, apply on the previous result
	else:
	new_idx = round(idx + tw["lookup"][idx])
	if not 0 <= new_idx < len(frame_range):
	# TODO: implementing this would need to actually have
	# retiming engine resolve process within AYON, resolving wraps
	# as curves, then projecting those into the previous media_range.
	raise NotImplementedError(
	"Unsupported consecutive timewarps (out of computed range)"
	)
	frame_range[idx] = frame_range[new_idx]
	frame_range[idx] = round(frame_number + tw["lookup"][idx] * time_scalar)
	continue
	# Consecutive timewarp, apply on the previous result
	new_idx = round(idx + tw["lookup"][idx])
	if 0 <= new_idx < len(frame_range):
	frame_range[idx] = frame_range[new_idx]
	continue
	# TODO: implementing this would need to actually have
	# retiming engine resolve process within AYON, resolving wraps
	# as curves, then projecting those into the previous media_range.
	raise NotImplementedError(
	"Unsupported consecutive timewarps (out of computed range)"
	)

If you preffer first raise exception then please use reversed condition if 0 > new_idx >= len(frame_range): (without not).

[iLLiCiTiT]

If time_scalar is int then this can be simplified to:

Suggested change

	frame_range = [media_in_trimmed]
	in_frame = media_in_trimmed + time_scalar
	while in_frame <= media_out_trimmed:
	frame_range.append(in_frame)
	in_frame += time_scalar
	frame_range = list(range(
	media_in_trimmed,
	media_out_trimmed + 1,
	time_scalar
	))

If is float then this is better.

[iLLiCiTiT]

Isn't this the same?

Suggested change

	filename, = tuple(collection)
	filename = collection[0]