DaalaTodo - Revision history

Jmspeex at 20:06, 18 February 2015

2015-02-18T20:06:14Z

← Older revision		Revision as of 20:06, 18 February 2015
Line 14:		Line 14:
	* Better MV split flag rate estimates		* Better MV split flag rate estimates
	* Does the 8-pixel MV offset influence quality in any way?		* Does the 8-pixel MV offset influence quality in any way?
			* Tuning for new lapping
			** quantization matrices
			** lambda
			** Haar DC

	== Known Broken/Suboptimal ==		== Known Broken/Suboptimal ==

Jmspeex at 22:25, 17 February 2015

2015-02-17T22:25:24Z

← Older revision		Revision as of 22:25, 17 February 2015
Line 20:		Line 20:
	** Never tuned on inter		** Never tuned on inter
	** Not considering chroma		** Not considering chroma
	* Code quantizers on a log scale
	* Re-order bitstream (e.g., don't code all MVs at the front, etc.)
	* Investigate bias introduced by coefficient scaling		* Investigate bias introduced by coefficient scaling
	* Take into account scaling differences from lapping on variable block size		* Take into account scaling differences from lapping on variable block size
	** Including Haar DC		** Including Haar DC
	* ~~Need~~ "skip higher bands" ~~flag~~		* Finish "skip higher bands" flags
	* Re-train zig-zags (particularly on inter)		* Re-train zig-zags (particularly on inter)
	* Implement RDO for "skip_rest" decisions.		* Implement RDO for "skip_rest" decisions.
	* Better RDO rates for gain/theta/noref		* Better RDO rates for gain/theta/noref
			* Don't code blocks that are outside the image

	== New Work ==		== New Work ==
Line 49:		Line 48:
	* Better encoding of the CfL sign		* Better encoding of the CfL sign
	* Add SIMD to the decoder search		* Add SIMD to the decoder search

			== Minor things ==

			* Code quantizers on a log scale

	== Visualization ==		== Visualization ==
Line 60:		Line 63:
	** Display prediction residual		** Display prediction residual

	== ~~Infrastructure~~ ==		== Bit-stream work (no expected quality improvement) ==

	* Need a bugzilla instance for issue tracking		* Need a bugzilla instance for issue tracking
			* Re-order bitstream (e.g., don't code all MVs at the front, etc.)

	== AreWeCompressedYet ==		== AreWeCompressedYet ==

MrZeus: /* Simple Things */ added request for Bugzilla!

2015-02-08T13:27:15Z

Simple Things: added request for Bugzilla!

← Older revision		Revision as of 13:27, 8 February 2015
Line 2:		Line 2:

	* Overflow checking for allocations in od_state_ref_imgs_init().		* Overflow checking for allocations in od_state_ref_imgs_init().
			* Implement [https://www.bugzilla.org/download/ Bugzilla] for Daala users to report bugs and enhancements.

	== Tuning ==		== Tuning ==

Jmspeex at 12:35, 8 February 2015

2015-02-08T12:35:31Z

← Older revision		Revision as of 12:35, 8 February 2015
Line 27:		Line 27:
	* Re-train zig-zags (particularly on inter)		* Re-train zig-zags (particularly on inter)
	* Implement RDO for "skip_rest" decisions.		* Implement RDO for "skip_rest" decisions.
			* Better RDO rates for gain/theta/noref

	== New Work ==		== New Work ==

Jmspeex at 12:37, 6 February 2015

2015-02-06T12:37:24Z

← Older revision		Revision as of 12:37, 6 February 2015
Line 26:		Line 26:
	* Need "skip higher bands" flag		* Need "skip higher bands" flag
	* Re-train zig-zags (particularly on inter)		* Re-train zig-zags (particularly on inter)
			* Implement RDO for "skip_rest" decisions.

	== New Work ==		== New Work ==

Jmspeex at 00:10, 6 February 2015

2015-02-06T00:10:44Z

← Older revision		Revision as of 00:10, 6 February 2015
Line 25:		Line 25:
	** Including Haar DC		** Including Haar DC
	* Need "skip higher bands" flag		* Need "skip higher bands" flag
			* Re-train zig-zags (particularly on inter)

	== New Work ==		== New Work ==

Jmspeex at 23:41, 5 February 2015

2015-02-05T23:41:43Z

← Older revision		Revision as of 23:41, 5 February 2015
Line 36:		Line 36:
	* SPIHT as k-tokenizer		* SPIHT as k-tokenizer
	* MV prediction clustering		* MV prediction clustering
			* Special case for K=2
			* Train Laplace tables

	== Entropy Coding ==		== Entropy Coding ==

Jmspeex at 12:00, 3 February 2015

2015-02-03T12:00:20Z

← Older revision		Revision as of 12:00, 3 February 2015
Line 12:		Line 12:
	* Better MC distortion metrics (SATD, some MSE/SATD hybrid, no-ref-aware, maybe table driven Theora-style, etc.)		* Better MC distortion metrics (SATD, some MSE/SATD hybrid, no-ref-aware, maybe table driven Theora-style, etc.)
	* Better MV split flag rate estimates		* Better MV split flag rate estimates
			* Does the 8-pixel MV offset influence quality in any way?

	== Known Broken/Suboptimal ==		== Known Broken/Suboptimal ==

MrZeus: /* AreWeCompressedYet */

2015-01-05T23:01:31Z

AreWeCompressedYet

← Older revision		Revision as of 23:01, 5 January 2015
Line 67:		Line 67:
	* Look at rd_tool options and improve		* Look at rd_tool options and improve
	* Add realtime constraint options for x265		* Add realtime constraint options for x265

			[[Category:Daala]]

Martin.leese: moved Daala To-do List to DaalaTodo: For consistency with other Todo pages

2014-12-25T16:52:38Z

moved Daala To-do List to DaalaTodo: For consistency with other Todo pages

← Older revision		Revision as of 16:52, 25 December 2014
(No difference)