HNM6 - Revision history

VAG: /* Normal decoding */

2012-10-13T14:55:27Z

Normal decoding

← Older revision		Revision as of 07:55, 13 October 2012
Line 209:		Line 209:
	[[Image:HNM6Spiral.png]]		[[Image:HNM6Spiral.png]]
	* Motion blocks are similar to WARP motion block, with minor enhancements:		* Motion blocks are similar to WARP motion block, with minor enhancements:
	** For keyframe motion is the same as for ~~WRAP~~, with the exception for blocks smaller than 4x4. For such blocks:		** For keyframe motion is the same as for WARP, with the exception for blocks smaller than 4x4. For such blocks:
	*** transformation mode = bits 12.14 of motion		*** transformation mode = bits 12.14 of motion
	*** xmotion = 63 - (bits 0..6 of motion)		*** xmotion = 63 - (bits 0..6 of motion)

VAG: /* Format modifications */

2012-10-13T14:53:25Z

Format modifications

← Older revision		Revision as of 07:53, 13 October 2012
Line 64:		Line 64:

	Each frame chunk followed by an audio chunk, however, this audio chunk size is not counted by frame's chunk size field.		Each frame chunk followed by an audio chunk, however, this audio chunk size is not counted by frame's chunk size field.
	~~Frame~~ chunk format:		Audio chunk format:

	u32 sig[4] -- "SOUN"		u32 sig[4] -- "SOUN"

VAG: /* File Format */

2012-10-13T14:52:22Z

File Format

← Older revision		Revision as of 07:52, 13 October 2012
Line 12:		Line 12:
	u8 audioflags -- nonzero value indicates file has APC sound (not authoritative), bits meaning:		u8 audioflags -- nonzero value indicates file has APC sound (not authoritative), bits meaning:
	7 : stereo		7 : stereo
	6..5 : frequency in units of ~~11025~~ (i.e. frequency = value * 11025)		6..5 : frequency in units of 11025Hz (i.e. frequency = value * 11025)
	u8 bpp		u8 bpp
	u16 width		u16 width

VAG: /* File Format */

2012-10-13T14:52:00Z

File Format

← Older revision		Revision as of 07:52, 13 October 2012
Line 11:		Line 11:
	u8 reserved[2] -- usually 0		u8 reserved[2] -- usually 0
	u8 audioflags -- nonzero value indicates file has APC sound (not authoritative), bits meaning:		u8 audioflags -- nonzero value indicates file has APC sound (not authoritative), bits meaning:
	8 : stereo		7 : stereo
	7..6 : frequency in units of 11025 (i.e. frequency = value * 11025)		6..5 : frequency in units of 11025 (i.e. frequency = value * 11025)
	u8 bpp		u8 bpp
	u16 width		u16 width

VAG: /* File Format */ explain several fields

2012-10-13T14:51:08Z

File Format: explain several fields

← Older revision		Revision as of 07:51, 13 October 2012
Line 10:		Line 10:
	u8 sig[4] -- file signature, "HNM6"		u8 sig[4] -- file signature, "HNM6"
	u8 reserved[2] -- usually 0		u8 reserved[2] -- usually 0
	u8 audioflags -- nonzero value indicates file has APC sound (not authoritative)		u8 audioflags -- nonzero value indicates file has APC sound (not authoritative), bits meaning:
			8 : stereo
			7..6 : frequency in units of 11025 (i.e. frequency = value * 11025)
	u8 bpp		u8 bpp
	u16 width		u16 width
Line 16:		Line 18:
	u32 filesize		u32 filesize
	u16 frames -- number of frames		u16 frames -- number of frames
	u16 ~~reserved2~~		u16 frames2 -- upper part of frame counter, ignored by some implementations
	u32 reserved3 -- usually 0, but sometimes "V107", "V108" - version?		u32 reserved3 -- usually 0, but sometimes "V107", "V108" - version?
	u16 speed -- playback speed in fps, may be zero (?assume 15 fps then?)		u16 speed -- playback speed in fps, may be zero (?assume 15 fps then?)

VAG: /* Video Format */ typos

2012-09-08T19:15:56Z

Video Format: typos

← Older revision		Revision as of 12:15, 8 September 2012
Line 69:		Line 69:

	== Video Format ==		== Video Format ==
	The video compression relies on a concept of Key-blocks, encoded with [[JPEG]]-like algorithm and motion blocks, derived from previously drawn blocks. The frame is encoded in 8x8 or smaller ~~block~~. The Key-blocks are always 8x8 and directly encoded, while motion blocks heavily rely on [[D&C]] approach and various block transformation techniques. Generally, this codec is quite similar to [[4XM]] codec and [[Mobiclip]] codecs, because it's written by the same people.		The video compression relies on a concept of Key-blocks, encoded with [[JPEG]]-like algorithm and motion blocks, derived from previously drawn blocks. The frame is encoded in 8x8 or smaller blocks. The Key-blocks are always 8x8 and directly encoded, while motion blocks heavily rely on [[D&C]] approach and various block transformation techniques. Generally, this codec is quite similar to [[4XM]] codec and [[Mobiclip]] codecs, because it's written by the same people.

	There are two variants of HNM6 video codec.		There are two variants of HNM6 video codec.
	* WARP codec. Prototype codec, simplified version. Used in a very little number of games.		* WARP codec. Prototype codec, simplified version. Used in a very little number of games circa 1997.
	* Normal (yes, that's how it's called internally) codec. Fully-featured version, widely used.		* Normal (yes, that's how it's called internally) codec. Fully-featured version, widely used.

Line 84:		Line 84:
	u32 jpegend -- offset of jpeg data buffer end		u32 jpegend -- offset of jpeg data buffer end

	All offsets a relative to this header.		All offsets are little-endian and relative to this header.

	* quality is a standard [[JPEG]] quality value (0..100%). Negative value is used to specify a keyframe (for Normal codec only.)		* quality is a standard [[JPEG]] quality value (0..100%). Negative value is used to specify a keyframe (for Normal codec only.)
Line 93:		Line 93:

	=== Key-blocks decoding ===		=== Key-blocks decoding ===
	Key-blocks are JPEG-encoded blocks in YUV 4:4:4 format. The requantization, [[IDCT]] and colorspace conversion steps are identical to standard [[JPEG]]. Standard zigzag (0, 1, 8, 16, 9, ...) and quantization tables (16, 11, 10, 16, 24, ... ; 17, 18, 24, 47, 99, ...) used as well.		Key-blocks are JPEG-encoded blocks in YUV 4:4:4 format. The requantization, [[IDCT]] and colorspace conversion steps are identical to standard [[JPEG]]. Standard zigzag (0,1,8,16,9,...) and quantization tables (16,11,10,16, 24,...; 17,18,24,47,99,...) used as well.

	The only custom feature used is the coefficients encoding. Instead of Huffman coding they are stored using RLE scheme.		The only custom feature used is the coefficients encoding. Instead of Huffman coding they are stored using RLE scheme.
Line 99:		Line 99:
	Macroblock's jpeg data is accessed by 4-bit nibbles, lower nibble of each byte first. Also, the following primitives are used:		Macroblock's jpeg data is accessed by 4-bit nibbles, lower nibble of each byte first. Also, the following primitives are used:
	* half - half of a nibble, upper part first, then lower (i.e. read a nibble, process upper part of it upon first use, lower part upon second)		* half - half of a nibble, upper part first, then lower (i.e. read a nibble, process upper part of it upon first use, lower part upon second)
	* s2 - signed two-bit 2's ~~compliment~~ value of a half, no zero point (i.e. 0b00 -> 1, 0b01 -> 2, 0b10 -> -2, 0b11 -> -1)		* s2 - signed two-bit 2's complement value of a half, no zero point (i.e. 0b00 -> 1, 0b01 -> 2, 0b10 -> -2, 0b11 -> -1)
	* s4 - signed four-bit 2's ~~compliment~~ value of a nibble, no zero point		* s4 - signed four-bit 2's complement value of a nibble, no zero point
	* s44 - signed eight-bit value of two nibbles, with zero point (this is basically (signextend(nibble1) << 4) \| nibble2)		* s44 - signed eight-bit value of two nibbles, with zero point (this is basically (signextend(nibble1) << 4) \| nibble2)

	For each plane of 8x8 coefficients, first ~~entry~~ is s44, then read nibbles and fill the rest of coefficients as follows until the whole plane is complete. Repeat for each Y, U and V plane.		For each plane of 8x8 coefficients, the first coefficient is s44, then read nibbles and fill the rest of coefficients as follows until the whole plane is complete. Repeat for each Y, U and V plane.

	0 - fill remainder of plane with zeros		0 - fill remainder of a plane with zeros
	1 - 0, 0, 0, 0		1 - 0, 0, 0, 0
	2 - 0, 1		2 - 0, 1
Line 141:		Line 141:


	* Keyblock is a JPEG-encoded macroblock		* Keyblock is a JPEG-encoded macroblock.
	* CrossCut means that the current block should be cut on to 4 smaller subblocks and each one processed independently, using the same scheme recursively.		* CrossCut means that the current block should be cut on to 4 smaller subblocks and each one processed independently, using the same scheme recursively.
	* Motion means that the block is motion-coded and should be copied from some other part of the frame. To do so, read the next entry of the motion buffer and process it as follows:		* Motion means that the block is motion-coded and should be copied from some other part of the frame. To do so, read the next entry of the motion buffer and process it as follows:
	** transformation mode = next 2 bits from bitbuffer (upper part) combined with bit 15 of motion (lower bit.)		** transformation mode = next 2 bits from bitbuffer (upper part) combined with bit 15 of motion (lower bit)
	** xmotion = 128 - (bits 7..14 of motion)		** xmotion = 128 - (bits 7..14 of motion)
	** ymotion = (if block is not 8x8, then 4) - (bits 0..6 of motion)		** ymotion = (if block is not 8x8, then 4) - (bits 0..6 of motion)
	* 2x2 blocks are always short motion coded, using no extra VLC		* 2x2 blocks are always short motion coded, using no extra VLC:
	** transformation mode = bits 1..3 of short motion		** transformation mode = bits 1..3 of short motion
	** motion index = bit 0 \|\| bits 4..7 \|\| bits 8..11 (total 9 bits)		** motion index = bit 0 \|\| bits 4..7 \|\| bits 8..11 (total 9 bits)
Line 161:		Line 161:
	ymotion = ymotion - 1		ymotion = ymotion - 1

	Block's source coordinates are the sum of current 8x8 macroblock's coordinates (not ~~it's~~ subblock's!) and the xmotion and ymotion respectively. If x-coordinate happens to go out of frame, it should be wrapped around the line. Copy block at the following coordinates to the current, applying the transformation as specified.		Block's source coordinates are the sum of current 8x8 macroblock's coordinates (not its subblock's!) and the xmotion and ymotion respectively. If x-coordinate happens to go out of frame, it should be wrapped around the line. Copy block at the following coordinates to the current, applying the transformation as specified.

	=== Normal decoding ===		=== Normal decoding ===
Line 203:		Line 203:
	* HorizontalCut splits the block horizontally, producing two subblocks of smaller height. Each one then processed recursively.		* HorizontalCut splits the block horizontally, producing two subblocks of smaller height. Each one then processed recursively.
	* VerticalCut is identical to HorizontalCut, but splits the block vertically.		* VerticalCut is identical to HorizontalCut, but splits the block vertically.
	* Skip simply copies the block from the previous frame		* Skip simply copies the block from the previous frame.
	* ShortMotion copies block from previous frame, using the spiral-encoded motion. Unlike regular motion, short motion is relative to subblock's coordinates, not the whole macroblock. Read ShortMotion index, then obtain block's relative coordinates as per illustration:		* ShortMotion copies block from previous frame, using the spiral-encoded motion. Unlike regular motion, short motion is relative to subblock's coordinates, not the whole macroblock. Read ShortMotion index, then obtain block's relative coordinates as per illustration:
	[[Image:HNM6Spiral.png]]		[[Image:HNM6Spiral.png]]
	* Motion blocks are similar to WARP motion block, with minor enhancements		* Motion blocks are similar to WARP motion block, with minor enhancements:
	** For keyframe motion is the same as for WRAP, with the exception for blocks smaller than 4x4. For such blocks:		** For keyframe motion is the same as for WRAP, with the exception for blocks smaller than 4x4. For such blocks:
	*** transformation mode = bits 12.14 of motion		*** transformation mode = bits 12.14 of motion
Line 224:		Line 224:
	*** xmotion = 31 - (bits 0..5 of motion)		*** xmotion = 31 - (bits 0..5 of motion)
	*** ymotion = 31 - (bits 6..11 of motion)		*** ymotion = 31 - (bits 6..11 of motion)
	** Don't forget to wrap x motion around the line if it goes out of frame		** Don't forget to wrap x motion around the line if it goes out of frame.

	=== Block transformation ===		=== Block transformation ===
Line 231:		Line 231:
	Copy block as is.		Copy block as is.
	==== 1 - Horizontal flip ====		==== 1 - Horizontal flip ====
	Swap left and right		Swap left and right:
	abc -> cba		abc -> cba
	==== 2 - Vertical flip ====		==== 2 - Vertical flip ====
	Swap up and down		Swap up and down:
	a c		a c
	b -> b		b -> b
	c a		c a
	==== 3 - Cross flip ====		==== 3 - Cross flip ====
	Swap both up-down and left-right		Swap both up-down and left-right:
	ab fe		ab fe
	cd -> dc		cd -> dc
	ef ba		ef ba
	==== 4 - Forward flip ====		==== 4 - Forward flip ====
	Swap block around /		Swap block around / :
	ab -> db		ab -> db
	cd ca		cd ca
	==== 5 - Forward rotate ====		==== 5 - Forward rotate ====
	Rotate block 90 degrees clockwise		Rotate block 90 degrees clockwise:
	ab eca fe		ab eca fe
	cd -> fdb -> dc		cd -> fdb -> dc
	ef ba		ef ba
	==== 6 - Backward rotate ====		==== 6 - Backward rotate ====
	Rotate block 90 degrees counterclockwise		Rotate block 90 degrees counterclockwise:
	ab bdf fe		ab bdf fe
	cd -> ace -> dc		cd -> ace -> dc
	ef ba		ef ba
	==== 7 - Backward flip ====		==== 7 - Backward flip ====
	Swap block around \		Swap block around \ :
	ab -> ac		ab -> ac
	cd bd		cd bd

VAG: /* Block transformation */

2012-09-08T18:53:24Z

Block transformation

← Older revision		Revision as of 11:53, 8 September 2012
Line 227:		Line 227:

	=== Block transformation ===		=== Block transformation ===
			During block copying, it can be additionally transformed. Transformation modes:
			==== 0 - Normal copy ====
			Copy block as is.
			==== 1 - Horizontal flip ====
			Swap left and right
			abc -> cba
			==== 2 - Vertical flip ====
			Swap up and down
			a c
			b -> b
			c a
			==== 3 - Cross flip ====
			Swap both up-down and left-right
			ab fe
			cd -> dc
			ef ba
			==== 4 - Forward flip ====
			Swap block around /
			ab -> db
			cd ca
			==== 5 - Forward rotate ====
			Rotate block 90 degrees clockwise
			ab eca fe
			cd -> fdb -> dc
			ef ba
			==== 6 - Backward rotate ====
			Rotate block 90 degrees counterclockwise
			ab bdf fe
			cd -> ace -> dc
			ef ba
			==== 7 - Backward flip ====
			Swap block around \
			ab -> ac
			cd bd

	=== Bitexact decoding ===		=== Bitexact decoding ===

VAG: realing

2012-09-08T18:32:29Z

realing

← Older revision		Revision as of 11:32, 8 September 2012
Line 92:		Line 92:
	* jpeg buffer points to array of encoded [[JPEG]] macroblocks		* jpeg buffer points to array of encoded [[JPEG]] macroblocks

	== ~~JPEG Blocks~~ decoding ==		=== Key-blocks decoding ===
	~~JPEG~~ blocks are encoded in YUV 4:4:4 format. The requantization, [[IDCT]] and colorspace conversion steps are identical to standard [[JPEG]]. Standard zigzag (0, 1, 8, 16, 9, ...) and quantization tables (16, 11, 10, 16, 24, ... ; 17, 18, 24, 47, 99, ...) used as well.		Key-blocks are JPEG-encoded blocks in YUV 4:4:4 format. The requantization, [[IDCT]] and colorspace conversion steps are identical to standard [[JPEG]]. Standard zigzag (0, 1, 8, 16, 9, ...) and quantization tables (16, 11, 10, 16, 24, ... ; 17, 18, 24, 47, 99, ...) used as well.

	The only custom feature used is the coefficients encoding. Instead of Huffman coding they are stored using RLE scheme.		The only custom feature used is the coefficients encoding. Instead of Huffman coding they are stored using RLE scheme.
Line 126:		Line 126:
	15 - s44		15 - s44

	== WARP decoding ==		=== WARP decoding ===
	Macroblocks are encoded using the following VLC codes:		Macroblocks are encoded using the following VLC codes:

Line 163:		Line 163:
	Block's source coordinates are the sum of current 8x8 macroblock's coordinates (not it's subblock's!) and the xmotion and ymotion respectively. If x-coordinate happens to go out of frame, it should be wrapped around the line. Copy block at the following coordinates to the current, applying the transformation as specified.		Block's source coordinates are the sum of current 8x8 macroblock's coordinates (not it's subblock's!) and the xmotion and ymotion respectively. If x-coordinate happens to go out of frame, it should be wrapped around the line. Copy block at the following coordinates to the current, applying the transformation as specified.

	== Normal decoding ==		=== Normal decoding ===
	Normal compression introduces interframes, more split modes and different short motion encoding.		Normal compression introduces interframes, more split modes and different short motion encoding.

Line 226:		Line 226:
	** Don't forget to wrap x motion around the line if it goes out of frame		** Don't forget to wrap x motion around the line if it goes out of frame

	== Bitexact decoding ==		=== Block transformation ===

			=== Bitexact decoding ===
	While you can use stock [[JPEG]] routines to decode Key-blocks, if you want to produce bitexact output you need to use very specific code to do so.		While you can use stock [[JPEG]] routines to decode Key-blocks, if you want to produce bitexact output you need to use very specific code to do so.

VAG: /* Normal decoding */ decoding

2012-09-08T18:26:33Z

Normal decoding: decoding

← Older revision		Revision as of 11:26, 8 September 2012
Line 164:		Line 164:

	== Normal decoding ==		== Normal decoding ==
			Normal compression introduces interframes, more split modes and different short motion encoding.

			Macroblocks are encoded using the following VLC codes:

			{\| border="1" cellpadding="5" style="border-collapse: collapse; border-style: dashed; border-color: #2f6fab;"
			\|- bgcolor="#f0f0f0" \|
			! Block type !! 8x8 !! 4x8 !! 8x4 !! 4x4 !! 2x4 !! 4x2 !! 2x2
			\|-
			\| colspan="8" style="text-align: center;" bgcolor="#f0f0f0" \| Keyframe
			\|-
			\| Keyblock \|\| 110 \|\| n/a \|\| n/a \|\| n/a \|\| n/a \|\| n/a \|\| n/a
			\|-
			\| HorizontalCut \|\| 00 \|\| 0 \|\| n/a \|\| 01 \|\| 1 \|\| n/a \|\| n/a
			\|-
			\| VerticalCut \|\| 01 \|\| n/a \|\| 0 \|\| 10 \|\| n/a \|\| 1 \|\| n/a
			\|-
			\| CrossCut \|\| 10 \|\| n/a \|\| n/a \|\| 11 \|\| n/a \|\| n/a \|\| n/a
			\|-
			\| Motion \|\| 111 \|\| 1 \|\| 1 \|\| 00 \|\| 0 \|\| 0 \|\| n/a
			\|-
			\| colspan="8" style="text-align: center;" bgcolor="#f0f0f0" \| Interframe
			\|-
			\| Keyblock \|\| 011 \|\| n/a \|\| n/a \|\| n/a \|\| n/a \|\| n/a \|\| n/a
			\|-
			\| HorizontalCut \|\| 100 \|\| 10 \|\| n/a \|\| 101 \|\| 111 \|\| n/a \|\| n/a
			\|-
			\| VerticalCut \|\| 101 \|\| n/a \|\| 10 \|\| 110 \|\| n/a \|\| 111 \|\| n/a
			\|-
			\| CrossCut \|\| 1110 \|\| n/a \|\| n/a \|\| 111 \|\| n/a \|\| n/a \|\| n/a
			\|-
			\| Skip \|\| 110 \|\| 11 \|\| 11 \|\| 100 \|\| 110 \|\| 110 \|\| 11
			\|-
			\| ShortMotion \|\| 00 \|\| 00 \|\| 00 \|\| 00 \|\| 0 \|\| 0 \|\| 0
			\|-
			\| Motion \|\| 010 \|\| 01 \|\| 01 \|\| 01 \|\| 10 \|\| 10 \|\| 10
			\|}

			* HorizontalCut splits the block horizontally, producing two subblocks of smaller height. Each one then processed recursively.
			* VerticalCut is identical to HorizontalCut, but splits the block vertically.
			* Skip simply copies the block from the previous frame
			* ShortMotion copies block from previous frame, using the spiral-encoded motion. Unlike regular motion, short motion is relative to subblock's coordinates, not the whole macroblock. Read ShortMotion index, then obtain block's relative coordinates as per illustration:
			[[Image:HNM6Spiral.png]]
			* Motion blocks are similar to WARP motion block, with minor enhancements
			** For keyframe motion is the same as for WRAP, with the exception for blocks smaller than 4x4. For such blocks:
			*** transformation mode = bits 12.14 of motion
			*** xmotion = 63 - (bits 0..6 of motion)
			*** ymotion = 6 - (bits 7..11 of motion)
			** Keyframe's 2x2 blocks are always motion-coded as above.

			** For interframe motion is different. For blocks 4x4 and larger:
			*** transformation mode = next 3 bits from bitbuffer
			*** if bit 15 of motion is zero then source is the previous frame, current frame otherwise
			*** xmotion = 128 - (bits 7..14 of motion)
			*** ymotion = (if from previous frame then 64, otherwise if block is not 8x8, then 4) - (bits 0..6 of motion)
			** For blocks smaller than 4x4:
			*** if bit 15 of motion is zero then source is the previous frame, current frame otherwise
			*** if from current frame - same as for keyframe, see above. for previous frame:
			*** transformation mode = next 3 bits from bitbuffer
			*** xmotion = 31 - (bits 0..5 of motion)
			*** ymotion = 31 - (bits 6..11 of motion)
			** Don't forget to wrap x motion around the line if it goes out of frame

	== Bitexact decoding ==		== Bitexact decoding ==

VAG: /* WARP decoding */ decoding

2012-09-08T15:48:43Z

WARP decoding: decoding

← Older revision		Revision as of 08:48, 8 September 2012
Line 127:		Line 127:

	== WARP decoding ==		== WARP decoding ==
			Macroblocks are encoded using the following VLC codes:

			{\| border="1" cellpadding="5" style="border-collapse: collapse; border-style: dashed; border-color: #2f6fab;"
			\|- bgcolor="#f0f0f0" \|
			! Block type !! 8x8 !! 4x4
			\|-
			\| Keyblock \|\| 11 \|\| n/a
			\|-
			\| Motion \|\| 10 \|\| 0
			\|-
			\| CrossCut \|\| 0 \|\| 1
			\|}


			* Keyblock is a JPEG-encoded macroblock
			* CrossCut means that the current block should be cut on to 4 smaller subblocks and each one processed independently, using the same scheme recursively.
			* Motion means that the block is motion-coded and should be copied from some other part of the frame. To do so, read the next entry of the motion buffer and process it as follows:
			** transformation mode = next 2 bits from bitbuffer (upper part) combined with bit 15 of motion (lower bit.)
			** xmotion = 128 - (bits 7..14 of motion)
			** ymotion = (if block is not 8x8, then 4) - (bits 0..6 of motion)
			* 2x2 blocks are always short motion coded, using no extra VLC
			** transformation mode = bits 1..3 of short motion
			** motion index = bit 0 \|\| bits 4..7 \|\| bits 8..11 (total 9 bits)
			** decode the index as follows:
			if index < 12 * 8
			xmotion = -2 - index % 12
			ymotion = 6 - index / 12
			else
			index -= 12 * 8
			xmotion = 19 - index % 32
			ymotion = -2 - index / 32
			if ymotion <= -8
			ymotion = ymotion - 1

			Block's source coordinates are the sum of current 8x8 macroblock's coordinates (not it's subblock's!) and the xmotion and ymotion respectively. If x-coordinate happens to go out of frame, it should be wrapped around the line. Copy block at the following coordinates to the current, applying the transformation as specified.

	== Normal decoding ==		== Normal decoding ==