This adds support for generating collide meshes when importing custom
models. A couple of things to keep in mind:
- A single `collide-mesh` may only have up to 255 vertices.
- When exporting a GLTF file in Blender, a `collide-mesh` will be
generated for every mesh object that has collision properties applied
(ideally, you would set all visual meshes to `ignore` and your collision
meshes to `invisible` in the OpenGOAL plugin's custom properties).
- Ensure that your actor using the model properly allocates enough
`collide-shape-prim-mesh`es for each `collide-mesh` ([example from the
original game that uses multiple
meshes](f6688659f2/goal_src/jak1/levels/finalboss/robotboss.gc (L2628-L2806))).
~One annoying problem that I haven't fully figured out yet (unrelated to
the actual functionality):
`collide-mesh`es are stored in art groups as an `(array collide-mesh)`
in the `art-joint-geo`'s `extra`, so I had to add a new `Res` type to
support this. The way that `array`s are stored in `res-lump`s is a bit
of a hack right now. The lump only stores a pointer to the array, so the
size of that is 4 bytes, but because we have to generate all the actual
array data too, the current `ResLump` code in C++ doesn't handle this
case well and would assert, so I decided to omit the asserts if an
`array` tag is present and "fake" the size so the object file is
generated more closely to how the game expects it until we figure out
something better.~
This was fixed by generating the array data beforehand and creating a
`ResRef` class that takes the pointer to the array data and adds it to
the lump.
This adds hfrag, but with a few remaining issues:
- The textures aren't animated. Instead, it just uses one texture.
- The texture filtering isn't as good as at it could be.
I also cleaned up a few issues with the background renderers:
- Cleaned up some stuff that is common to hfrag, tie, tfrag, shrub
- Moved time-of-day color packing stuff to FR3 creation, rather than at
level load. This appears to reduce the frame time spikes when a level is
first drawn by about 5 or 6 ms in big levels.
- Cleaned up the x86 specific stuff used in time of day. Now there's
only one place where we have an `ifdef`, rather than spreading it all
over the rendering code.
This fixes issues with certain Jak 3 levels not rendering because there
is a mismatch between the DGO name, nickname and real level name (bsp
name).
FR3s use a different filename, so you can delete the ones you have after
this is merged.
This affects custom levels, but I don't have that toolchain set up so
someone else will have to test that.
Added framework to do texture animations entirely in C++. Currently only
works on relatively simple ones, and doesn't handle updating all
parameters - only the speeds.
Connected texture animations to merc and tfrag for skull gems, dark
bomb, and scrolling conveyors.
Cleaned up Tfragment/Tfrag3, which used to be two classes. This was one
of the first C++ renderers, so it had a weird design.
I havn't tested it yet, but I can almost guarantee that atleast `goalc`
will not work in the slightest!
But the project is atleast fully compiling. My hope is to start
translating some AVX to NEON next / get `goalc` working...eventually.
Trying to make up for some of the startup speed lost in the SDL
transition. This saves about 1s from start (from ~3s), and about 500 MB
of RAM.
- Faster TIE unpack by merging matrix groups, more efficient vertex
transforms, and skipping normal transforms on groups with no normals.
- Refactor generic merc and merc to use a single renderer with multiple
interfaces, rather than many renderers. Removed "LightningRenderer" as a
special thing, but Warp is still special
- Add more profiling stuff to startup and the loader.
- Remove `SDL_INIT_HAPTIC` - this turned out to be needed for
force-feedback steering wheels, and not needed for controller vibration
- Switched `vag-player` to use quicksort instead of the default GOAL
sort (very slow)
This moves the blerc math from mips2c to the Merc2 renderer, and uses
floats instead.
We could potentially do this on the GPU, which would be even faster, but
this isn't that slow in the first place.
Saves 16 bits and lets us align the `color_index` field properly.
This shouldn't improve or decrease performance by any noticeable amount
except maybe in really low end systems.
Definitely needs a clean up pass, but I think the functionality is very
close.
There's a few "hacks" still:
- I am using the emerc logic for environment mapping, which doesn't care
about the length of the normals. I can't figure out how the normal
scaling worked in etie. I want to do a little bit more experimentation
with this before merging.
- There is some part about adgifs for TIE and ETIE that I don't
understand. The clearly correct behavior of TIE/ETIE is that the alpha
settings from the adgif shader are overwritten by the settings from the
renderer. But I can't figure out how this happens in all cases.
- Fade out is completely disabled. I think this is fine because the
performance difference isn't bad. But if you are comparing screenshots
with PCSX2, it will make things look a tiny bit different.
Three main changes:
- Adds support for the texture scrolling effect used on conveyor belts,
and turn it on for jak 2.
- Use merc instead of generic in jak 1 for ripple/water/texscroll stuff
(non-ocean water, lava, dark eco, etc). This is a pretty big speedup in
a lot of places.
- Fix a really old bug with blending mode used to draw environment maps.
The effect is that envmaps were half as bright as they should have been.
As usual, there's a flag to go back to the old behavior on jak 1. Set
these to `#t` to use generic like we used to.
```
*texscroll-force-generic*
*ripple-force-generic*
```
The format has changed, and everything must be rebuilt (C++, FR3's, GOAL
code)
Support rendering eyes with merc for both jak 1 and jak 2.
For jak 1, everything should look the same, but merc will be used to
draw eyes. This means that jak is now fully drawn with merc!
For jak 2, eyes should draw, but there are still a few issues:
- the tbp/clut ptr trick is used a lot for these eye textures, so
there's a lot that use the wrong texture
- I had to enable a bunch more "texture uploads" (basically emulating
the ps2 texture system) in order to get the eyes to upload. It would be
much better if the eye renderer could somehow grab the texture from the
merc model info, skipping the vram addressing stuff entirely. I plan to
return to this.
- I disabled some sky draws in `sky-tng`. After turning on pris2
uploads, the sky flashed in a really annoying way.
This PR adds a feature to merc2 to update vertices. This will be needed
to efficient do effects like blerc/ripple/texture scroll. It's enabled
for blerc in jak 1 and jak 2, but with a few disclaimers:
- currently we still use the mips2c blerc implementation, which is slow
and has some "jittering" because of integer precision. When porting to
PC, there was an additional synchronization problem because blerc
overwrites the merc data as its being read by the renderers. I _think_
this wasn't an issue on PS2 because the blerc dma is higher priority
than the VIF1 DMA, but I'm not certain. Either way, I had to add a mutex
for this on PC to avoid very slight flickering/gaps. This isn't ideal
for performance, but still beats generic by a significant amount in
every place I tested. If you see merc taking 2ms to draw, it is likely
because it is stuck waiting on blerc to finish. This will go away once
blerc itself is ported to C++.
- in jak 1, we end up using generic in some cases where we could use
merc. In particular maia in village3 hut. This will be fixed later once
we can use merc in more places. I don't want to mess with the
merc/generic selection logic when we're hopefully going to get rid of it
soon.
- There is no support for ripple or texture scroll. These use generic on
jak 1, and remain broken on jak 2.
- Like with `emerc`, jak 1 has a toggle to go back to the old behavior
`*blerc-hack*`.
- In most cases, toggling this causes no visual differences. One
exception is Gol's teeth. I believe this is caused by texture coordinate
rounding issues, where generic has an additional float -> int -> float
compared to PC merc. It is very hard to notice so I'm not going to worry
about it.
This adds environment mapping support to `Merc2`, and turns it on for
Jak 1 and Jak 2.
- The performance is much better
- Jak 1 can be toggled back to the old behavior with `(set! *emerc-hack*
#f)`. The new environment mapping is identical to the old one everywhere
I checked.
- Jak 1 still falls back to generic for ripple/texscroll/blerc/eyes -
there's still no dynamic texture or vertex updating support. The eye
detection stuff will sometimes flag stuff as eyes which is not eyes,
which is fine, but means that generic will be used in some places where
emerc could be used. For example, the shiny plates on jak's arm will be
drawn with generic because jak has eyes.
- Jak 2 hasn't been checked super carefully against PCSX2 yet.
- Jak 2 still isn't technically using emerc, but instead putting emerc
models in the merc bucket.
- The interface to merc is a lot different now and totally custom
OpenGOAL DMA code. The original merc drawing asm doesn't run anymore.
- The FR3 format changed
- Something funky going on with foreground lighting in escape, but
doesn't seem to be related to this change?
Performance comparison, jak 1, in likely the most generic-merc heavy
spot:
* wip
* learning about colors
* gltf node stuff working
* cleanup
* support textures
* bvh generation seems reasonable
* tree layout
* frag packer, untested and doesnt do real stripping yet
* temp
* working collide frags
* handle bad inputs better
* clean up
* format
* include
* another include
* reorganize for release build use
* tfrag3 data for merc2
* dma hooks for merc2
* start designing merc2 opengl, seems like the simple approach will be the best here
* before bone packing experiment
* fix up bones.gc
* use uniform buffer
* speedup, fix faces and eyes
* final fixes
* ci: fix windows releases (hopefully)
* scripts: fix Taskfile file references for linux
* asserts: add `ASSERT_MSG` macro and ensure `stdout` is flushed before `abort`ing
* asserts: refactor all `assert(false);` with a preceeding message instances
* lint: format
* temp...
* fix compiler errors
* assert: allow for string literals in `ASSERT_MSG`
* lint: formatting
* revert temp change for testing
* first pass
* first pass at shrinking fr3s
* only need to load vertices once
* avx2 detect and switch
* fix build
* another ifx'
* one more
* fix the sky and stupid math bug in size check
* MSAA
* support different TIE lods
* tfrag lod picking
* always compress screenshots
* mute annoying nvidia opengl message
* use a string here instead
* oops
* more mistakes...
* temp
* temp
* wip
* more progress on the instance asm
* first half of tie extraction, up to dma lists
* more tie extraction
* first part figured out maybe
* bp1 loop seems to work, bp2 loop does not
* bp1 and bp2 appear working. sadly ip is needed
* ip1 outline, not working ip2
* just kidding, ip2 seems to work
* extraction seems to work
* basic rendering working
* tie fixes
* performance optimization of tie renderer
* hook up tie to engine
* fix more bugs
* cleanup and perf improvements
* fix tests
* ref tests
* mm256i for gcc
* CLANG
* windows
* more compile fixes
* fix fast time of day
* small fixes
* fix after merge
* clang
* begin work
* work
* working objs
* exporting
* it works
* before some time of day fixes
* add time of day interp and also fix zbuffer
* some small blending fixes
* improve randomess
* clean up extraction and missing blend mode
* culling, time of day, more level fixes
* more cleanup
* cleanup memory usage
* windows fix
