Commit Graph

93 Commits

Author SHA1 Message Date
kris
97403ab431 Create data dir before writing to it 2023-01-25 22:57:18 +00:00
kris
53dac6a47c Disable debugging assertions now they've served their purpose 2023-01-24 21:49:33 +00:00
kris
6a8d49bd97 - Stop using the 5x weighting for diff values, it was preventing from
consistently finding enough additional offsets (~2.5x avg instead of
  >2.9)

- Remove instrumentation for fill rate now that it's served its purpose
2023-01-24 21:49:33 +00:00
kris
990e1c9d74 - have Bitmap.apply() update the memory representation instead of
requiring callers to keep track of it

- stop trying to cache content_deltas, I think it results in losing
  deltas.  Instead just recompute the deltas for each page as we need
  it.  This is fast enough in practice.

- track the average fill rate for the additional offsets we emit.
  This should be close to 3 if we're succeeding in finding enough
  collateral work

- overhaul how we pass in the target memory maps.  The previous way
  didn't make sense: we weren't actually encoding for the target video
  frame, but were using an inconsistent mix of old and new frames.  I
  think this was causing image artifacting because we were aiming for
  the wrong thing.

- Add some debugging assertions that were used to track this down.
2023-01-24 21:49:33 +00:00
kris
6b612ffb0a Normalize audio at 0.5/99.5%iles to clip less 2023-01-24 21:49:33 +00:00
kris
1d5bcfd74e Optimize make_data_tables and use numpy.save instead of pickling. The
file sizes are a bit larger but it unblocks updating to python 3.8.
2023-01-24 21:49:33 +00:00
kris
7384878ecb Disable debugging assertions now they've served their purpose 2023-01-17 21:44:43 +00:00
kris
f7f5950637 - Stop using the 5x weighting for diff values, it was preventing from
consistently finding enough additional offsets (~2.5x avg instead of
  >2.9)

- Remove instrumentation for fill rate now that it's served its purpose
2023-01-17 21:42:28 +00:00
kris
0a3c81c8c6 - have Bitmap.apply() update the memory representation instead of
requiring callers to keep track of it

- stop trying to cache content_deltas, I think it results in losing
  deltas.  Instead just recompute the deltas for each page as we need
  it.  This is fast enough in practice.

- track the average fill rate for the additional offsets we emit.
  This should be close to 3 if we're succeeding in finding enough
  collateral work

- overhaul how we pass in the target memory maps.  The previous way
  didn't make sense: we weren't actually encoding for the target video
  frame, but were using an inconsistent mix of old and new frames.  I
  think this was causing image artifacting because we were aiming for
  the wrong thing.

- Add some debugging assertions that were used to track this down.
2023-01-17 21:39:05 +00:00
kris
5c728e7ff1 Allow edit transposes again, there doesn't seem to be a good reason to prevent it and I'm not sure it can even happen in practice 2023-01-17 21:10:49 +00:00
kris
efe821d215 Fix warning 2023-01-17 21:09:34 +00:00
kris
b1132e97f1 Remove pregenerated data files, they are using up all my LFS quota 2023-01-17 21:09:18 +00:00
kris
4f6d7a796a Fix some numpy deprecation warnings 2023-01-17 12:26:25 +00:00
kris
451523bdef Support for custom output bitrate, e.g. to target //gs in 2.8MHz mode.
For some reason playback speed is only about 1.6x (probably due to
slowing down accesses to the I/O page to 1MHz, so as not to mess up
hardware timings), but happily this comes within 3% of being 44100/2.
2019-07-14 22:05:20 +01:00
kris
ce3bed0d38 Try resampling with scipy and scaling audio power 2019-07-13 14:32:21 +01:00
kris
34f2663448 Initial checkin of edit distance matrices for HGR/DHGR palettes. 2019-07-11 23:44:32 +01:00
kris
1b20cb8f63 Remove obsolete files 2019-07-11 23:43:59 +01:00
kris
f3d03a1b87 Update comments and fix some bugs
make_edit_distance
- use MASKED_DOTS since it does not have a simple relationship to the
  HEADER_BITS/BODY_BITS for HGR
- try disabling transposition distances for Damerau-Levenshtein, this
  may give better quality

screen
- introduce separate notion of MASKED_DOTS which is the number of
  (coloured) pixels we can extract from MASKED_BITS.  For HGR this is
  not the same.
- fix bug in _fix_array_neighbours that was not fixing headers for HGR
- don't cache everything in byte_pair_differences, it's effectively
  unbounded.  Using 1M for LRU size seems to work just as well in
  practise, without leaking memory.
- fix bug in _diff_weights when comparing content, we want to evaluate
  the effect of storing content byte in each offset separately, not
  cumulatively.
- add a consistency check function (not currently wired up) to assert
  that headers/footers are in sync across columns
- HGR should have 16 body bits, this was causing headers not to
  propagate correctly to/from neighbouring column
- add test case for this bug

video
- Use 8 random bits consistently, using 16 in some places may have
  introduced bias
- ignore palette bit when comparing 0x00 and 0x7f in sanity check
2019-07-11 23:40:00 +01:00
kris
b05c3bec1e - Add a progress bar while computing the edit distance matrices
- Only compute the upper triangle since the matrix is symmetrical and
  we can reconstruct the rest cheaply at load time

- Compute edit distances for HGR as well by making use of the fact that
  the masked representation is 14-bit but still sparse.

- Add sanity checks for various invariants of the edit distance matrices.
2019-07-07 21:25:07 +01:00
kris
ab29b01d0f Finish implementing HGRBitmap support.
- For HGRBitmap introduce a packed representation in the form
  hhHaaaaaaaABbbbbbbbFff where capitals indicate the location of the
  palette bit.  i.e. for header and footer we include the neighbouring
  2 data bits as in DHGR but also the palette bit from that byte, which
  is necessary to understand how these data bits unpack into dots.

  The nonstandard ordering of the palette bit for the odd data byte (B)
  is so that the masking by byte offset produces a contiguous sequence
  of bits, i.e. the 14-bit masked representation is still dense.

- Introduce a to_dots() classmethod that converts from the masked bit
  representation of dots influenced by a screen byte to the actual
  sequence of screen dots.  For DHGR this is the identity map since
  there are no shenanigans with palette bits causing dots to shift
  around.

- Add a bunch more unit tests, and add back the Sather tests for HGR
  artifact colours from palette bit interference, which now all pass!

- Reduce the size of the precomputed edit distance matrix by half by
  exploiting the fact that it is symmetrical under
  i << N + j <-> j << N + i where N is the size of the masked bit
  representation (i.e. transposing the original (i, j) -> dist
  metric matrix).
2019-07-07 21:22:44 +01:00
kris
16c4faa66d Support decoding arbitrary number of dots as nominal colour sequence.
This is needed for HGR support.
2019-07-07 21:13:28 +01:00
kris
fc4a63fffe Support HGR mode again 2019-07-07 21:12:10 +01:00
kris
740dafbd74 Remove commented out code 2019-07-04 15:54:05 +01:00
kris
5c550d8524 Separate the details of the bitmap packing from operations on the
packed representation (diff, apply etc).  This allows the (D)HGRBitmap
classes to focus on the bitmap packing and share common logic.

Numpy has unfortunate long-standing bugs to do with type coercion of
np.uint64, which leads to spurious "incompatible type" warnings when
e.g. operating on a np.uint64 and some other integer type.  To work
around this we cast explicitly to np.uint64 everywhere.

Get tests working again - for now HGR tests in screen_test.py are
disabled until I finish implementing new packing.

HGRBitmap is still incomplete although closer.
2019-07-04 15:21:20 +01:00
kris
666272a8fc Checkpoint WIP for easier comparison to dhgr branch:
- naive version of NTSC artifacting, it uses a sliding 4-bit window to
  assign a nominal (D)HGR colour to each dot position.  A more
  sophisticated/correct implementation would model the YIQ signal
  directly.

- Switch DHGRBitmap implementation to use a 34-bit representation of
  the 4-byte tuple, comprised of a 3-bit header and footer, plus
  4*7=28-bit body.  The headers/footers account for the influence on
  neighbouring tuples from the 4-bit NTSC window.

- With this model each screen byte influences 13 pixels, so we need to
  precompute 2^26 edit distances for all possible (source, target)
  13-bit sequences.

- Checkpointing not-yet-working HGR implementation.

- Add new unit tests but not yet all passing due to refactoring
2019-07-02 22:40:50 +01:00
kris
b2c00784b0 Minor code cleanups 2019-06-21 22:08:22 +01:00
kris
7d041a6a84 Store video frames in {basename}/{mode}/{palette}/ directory to
disambiguate different values (for palettes in particular)
2019-06-20 21:15:52 +01:00
kris
a732a0e034 Fix merge 2019-06-19 22:28:31 +01:00
kris
e3cb6fe7d1 Merge branch 'dhgr' of https://github.com/KrisKennaway/ii-vision into dhgr 2019-06-19 22:16:36 +01:00
kris
2cda5ea22c Merge branch 'dhgr' of https://github.com/KrisKennaway/ii-vision into dhgr 2019-06-19 22:14:24 +01:00
kris
cf3c687a8f Fix merges after rebase 2019-06-19 22:10:15 +01:00
kris
4d9e1e510b random.getrandbits() is much faster than random.randint() 2019-06-19 22:06:54 +01:00
kris
edefe649f4 Parametrize the RGB palette to encode with, and support both NTSC and
IIGS palettes.

Move the palette diff_matrix generation into make_data_tables.py since
that is the only place it is used.

Demand-load the edit distance matrices when transcoding.
2019-06-19 22:06:54 +01:00
kris
696eb61bf4 Compress the precomputed edit_distance arrays 2019-06-19 22:04:06 +01:00
kris
f07cabafb7 Add type annotations 2019-06-19 22:04:02 +01:00
kris
33aa4d46c4 Rename FrameSequencer to FrameGrabber and break out into separate file.
Add a test case that the bmp2dhr output of input filenames containing
'.'  are handled correctly.

Break out video.Mode into video_mode.VideoMode to resolve circular
dependency.
2019-06-19 22:04:02 +01:00
kris
fac2089206 Add tests for _diff_weights 2019-06-19 22:02:13 +01:00
kris
5d9ffe7d8b No longer used 2019-06-19 22:02:13 +01:00
kris
fd49736b71 FrameSequencer:
- Extract out a (File)FrameSequencer class from Video to encapsulate
the generation of still frames.  This also makes Video easier to test.

- Fix FileFrameSequencer.frames() to correctly handle filenames
  containing '.'

- Temporarily switch to the BMP2DHR NTSC palette (#5) for evaluation.

Video:
- Temporarily hardcode DHGR decoding

- Optimize _heapify_priorities() by using numpy to vectorize the
  construction of the list of tuples.  This requires changing the
  random nonce to an int so the intermediate array has a uniform type.

- Use the efficient 28-bit representation of DHGR (aux, main, aux,
  main) tuples introduced in DHGRBitmap to evaluate diffs

- Switch to np.int type for accumulating diffs, and random.randint(0,
  10000) instead of float for nonce values.

- Fix/improve some of the error evaluation in _index_changes:
  - skip offsets whose diffs have already been cleared
  - hoist some stuff out of _compute_error into the parent

- Add some validation that when we run out of work to do with a frame,
  the source and target memory maps should be equal.  This isn't
  happening sometimes, i.e. there is a bug.
2019-06-19 22:02:13 +01:00
kris
15c77f2465 Remove unused code.
Add a new DHGRBitmap class that efficiently represents the
DHGR interleaving of the (aux, main) MemoryMap as a sequence of
28-bit integers.

This allows for easily extracting the 8-bit and 12-bit subsequences
representing the DHGR pixels that are influenced when storing a byte
at offsets 0..3 within the interleaved (aux, main, aux, main)
sequence.

Since we have precomputed all of the pairwise differences between
these 8- and 12-bit values, this allows us to efficiently compute the
edit distances between pairs of screen bytes (and/or arrays)
2019-06-19 21:54:52 +01:00
kris
9c90665e96 Fix warnings 2019-06-19 21:54:52 +01:00
kris
494503aa88 Precompute the edit distance between the 8-bit (2-pixel) and 12-bit
(3-pixel) sequences that may be modified when storing bytes to the
DHGR display.

This relies on producing an efficient linear representation of the
DHGR framebuffer in terms of a packed 28-bit representation of (Aux,
Main, Aux, Main) screen bytes.
2019-06-19 21:54:52 +01:00
kris
b33b730bfa Map DHGR colours to RGB values (for now using the BMP2DHR default
NTSC palette), and compute the matrix of pairwise perceptual
differences between these colours, using CIE2000 delta values.
2019-06-19 21:54:49 +01:00
kris
cb287ea0e7 Add a DHGRColours enum mapping the DHGR logical colours to their
(memory-order) 4-bit values.
2019-06-19 21:54:49 +01:00
kris
5dcbca4aae Exit gracefully when running out of video frames 2019-06-19 21:54:07 +01:00
kris
38663088ef Remove CycleCounter, it's unused 2019-06-19 21:51:45 +01:00
kris
bd0aa6ad6b Simplify the accounting of audio ticks in video framing, and fix an off-by-one that was causing audio and video to become desynchronized.
Switch from 14340Hz to 14700Hz, which is 44100/3 i.e. a divisor of the
most common audio input sample rate.  This gives better quality
(and/or faster to encode) audio quality at the cost of producing <2%
faster playback.
2019-06-19 21:47:49 +01:00
kris
32bb4347ad Fix unit test 2019-06-19 21:45:58 +01:00
kris
a6d7cf73b1 Fix warning about missing newline at EOF 2019-06-17 23:46:52 +01:00
kris
079036042a Add TODO about Opcode.apply being currently unused 2019-06-17 23:46:24 +01:00