Less than 100 instructions left to implement, with the vast vast
majority being load-and-stores. These will likely be knocked out quickly
but they require a more involved implementation than just simply
translating the instructions (several need multiple instructions, others
may need reserved registers (x16 or x17 are common for this purpose))
This is a good milestone to get something pushed to master.
Three changes:
1. Add lazy merging/replacing for texture `rgba_bytes`.
2. Use a single thread when decompiling with texture replacements.
3. Drop textures after serializing and before compression.
These changes should enable users to install **massive** texture packs
without issue.
### The Problem:
The tpages list at the top of the JSONC, never actually gets used or is
written into memory:
<img width="270" height="195" alt="1"
src="https://github.com/user-attachments/assets/d3983363-23db-496e-bcea-136ab0e04058"
/>
This means that previously, the only way that texture remapping was
working was in very specific circumstances, which is when the autofill
would trigger. The autofill would only trigger under these specific
conditions:
- Both a sky remap and level remap are used
- The sky remap and level remap are the same level
- The tpages list is empty
If you went against any of these three parameters, texture remapping
could not be done.
### The Solution:
I simply reversed the logging line
[here](https://github.com/open-goal/jak-project/blob/0fa93ce7b83d1621f3fd040625bfa865eadaeef0/goalc/build_level/jak1/build_level.cpp#L240),
which turns the tpages into numbers in order to print them, and reversed
the shift from << to >> to instead take the numbers and write them into
memory.
```cpp
if (!tpages.empty()) {
file.texture_ids.resize(tpages.size());
for (size_t i = 0; i < tpages.size(); ++i) {
file.texture_ids[i] = tpages[i] << 20;
}
}
```
Pictured: A sunken level remap (correct generic textures on water) with
a village1 sky (so not depending on the autofill that only runs when the
sky and level are matching):
<img width="1716" height="1012" alt="aaa"
src="https://github.com/user-attachments/assets/1af3eddb-a5fb-4387-b644-bf84ba60ab62"
/>
After verifying so many things about the BVH trees being correct, I
noticed a pattern across missing collision: all missing collision was
always a face-containing child that was inside a node alongside another
non-face-containing parent.
This minimalist fix simply adds a wrapper so that when a face-containing
child exists alongside a non-face-containing parent, ie. `if (has_leaves
&& has_not_leaves)`, then it simply wraps the face-containing node in
another parent, so that face-containing spheres and non-face-containing
parents don't co-exist in the same node.
The result: restored collision for missing sections in custom levels.
If long_name is too long, it will result in the level being invisible
but the collide loading, lets detect this and point the user in correct
direction to solve it with a clear error message.
---------
Co-authored-by: Hat Kid <6624576+Hat-Kid@users.noreply.github.com>
I discovered that `yakow`s are kinda broken if you try to use them in
custom levels in jak 2/3.
It's due to the missing `yakow-lod0` texture and associated fix,
replacing it with `yak-medfur-end`. This solution works fine for the
decompiler on vanilla levels.
But for building custom levels, the requested art-groups were being
handled before the textures were, and so it was impossible to have
`yak-medfur-end` on hand to do the replacement. You'd hit an exception
here because `idx_in_level_texture` would still be `INT32_MAX`:
https://github.com/open-goal/jak-project/blob/c08118509b84feba002bd9e208f49162b4218556/decompiler/level_extractor/extract_merc.cpp#L806
My fix was just to swap the order when building custom levels, and
handle the textures first. I only made the changes for jak2/3, because I
see @Hat-Kid has a slightly different implementation for jak1.
There's one other small change relating to the `combo_id` /
`pc_combo_tex_id` short-circtuiting - I think `pc_combo_tex_id` is
always 0 for vanilla textures? So initially `yakow-lod0` actually ended
up matching the `combo_id` of the checkerboard texture from the
test-zone GLB. I just added another sanity check here that the texture
names match too.
(I also added yakows in the test-zone.jsonc files 🐄)
Follow-up later on improving the validation of the extractor (that may
require testing with all iso versions), but this is just some simple
safe-guarding code that will improve the error messaging, instead of
just a vague invalid map error.
Custom levels for Jak 2/3 now support envmapped TIE geometry. The TIE
extract was also changed to ignore materials that have the specular flag
set, but are missing a roughness texture.
Jak 2/3 now also support the `build-actor` tool.
The `build-custom-level` and `build-actor` macros now have a few new
options:
- Both now have a `force-run` option (`#f` by default) that, when set to
`#t`, will always run level/art group generation even if the output
files are up to date.
- `build-custom-level` has a `gen-fr3` option (`#t` by default) that,
when set to `#f`, will skip generating the FR3 file for the custom level
and only generate the GOAL level file to skip the potentially slow
process of finding and adding art groups and textures. Useful for when
you want to temporarily edit only the GOAL side of the level (such as
entity placement, etc.).
- `build-actor` has a `texture-bucket` option (default 0) which will
determine what DMA sink group the model will be placed in, which is
useful to determine the draw order of the model. Previously, this was
omitted, resulting in shadows not drawing over custom actors because the
actors were put in a bucket that is drawn after shadows (this behavior
can be restored with `:texture-bucket #f`).
A few improvements to color palette selection. These were made by
tracing some particularly bad colors through and seeing where it made
obviously bad decisions for splitting. I tested on crystal cave, and a
test GLBs from Kuitar that previously had issues with alpha.
- The previous approach to splitting was based on trying to keep a tree
of deduplicated colors balanced (same count in each leaf). This is not
really a good idea for generating color palettes. A better approach is
to try to minimize the volume of the child node, limiting how inaccurate
a color can be. Splitting is now chosen based on the average of the
_deduplicated channel values_, which in practice seems to work pretty
well for Kuitar's levels. Other approaches could work here too.
- The previous approach of alternating through dimensions to split on
was kept.
- The depth of the KD tree during the initial split was increased,
allowing it to use up to 8192 colors, instead of just 1024.
- In most cases, not all child nodes of the tree have colors in them,
meaning that a tree of depth 13 would have less than 8192 colors. If
this happens, child nodes are split until the color count reaches 8192.
The selection of which nodes to split is somewhat arbitrary, but is
breadth-first. The axis for splitting is the one with the largest range.
(which might be a better idea in general?)
On crystal-cave, the worst case color error was reduced from 221 to 9.
---------
Co-authored-by: water111 <awaterford1111445@gmail.com>
Store a small database of which models have already been swapped out in
a level to prevent duplicate processing.
Also a small fix for cases where using a model replacement that has no
armature would cause merc nightmares due to only having a `max_bones` of
3.
Base implementation of the popup menu and speedrunner mode in Jak 3.
Autosplitter is untested because I'm on Linux.
Also a couple of other misc changes:
- Model replacements can now have custom bone weights. Needs the "Use
Custom Bone Weights" property (provided by the OpenGOAL Blender plugin)
enabled in Blender.
- Better error message for lump syntax errors in custom level JSON
files.
By adding the `draco` library as a dependency, `tinygltf` can support
GLB files compressed with the Draco compression algorithm which allows
for drastically reduced file sizes for custom levels (TFL's Crescent Top
GLB for example went from 135 MB to 37 MB).
- Bug fix to KD tree splitting, should fix cases with bad vertex
colors/alphas.
- Normalize normals instead of asserting if they are the wrong length.
**the fact that blender exports normals incorrectly is a bug and I doubt
their implementation is correct if you've scaled things on only on
axis.**
- Automatically resize metallic texture (envmap strength) if it doesn't
match the size of the rgb texture instead of asserting
Co-authored-by: water111 <awaterford1111445@gmail.com>
This adds some new JSON entries to custom levels so they can support
vanilla sky textures and the texture remapping tables, allowing for
proper textures on objects that use `generic`, like dark eco pools or
dying enemies.
The comments explain it in more detail, but the gist is:
For skies:
- `sky` needs to be a vanilla level that has sky textures.
- The alpha tpage (fourth entry in `tpages`) needs to be that vanilla
level's alpha tpage (if `tex_remap` is the same level as `sky`, this
will be handled automatically).
- The tpage needs to be added to the custom level `.gd` and to
`textures` in the JSON.
- In `level-info.gc`, `sky` needs to be `#t`, your level's mood needs to
call `update-mood-sky-texture` (the default mood, `update-mood-default`,
handles this as an example) and `sun-fade` needs to be nonzero for the
sun to show up.
For `generic` textures:
- `tex_remap` needs to be the name of a vanilla level.
- When using a vanilla level's remap table, you need to adhere to the
order of the files in that level's `.gd` in your own level.
- Code files are first.
- Then the tpages (in the order `tfrag`, `pris`, `shrub`, `alpha`,
`water`).
- Then the art groups.
- Lastly, the level file.
- The tpages need to be added to the `textures` in the JSON.
This only applies to the background for now:
- support for alpha for vertex colors in custom levels
- switch time of day palette generation from octree to k-d tree
- support for alpha masking in custom levels
- support for transparent textures
- support for envmap in custom levels
---------
Co-authored-by: water111 <awaterford1111445@gmail.com>
This adds a feature to `build_actor` to support importing skeletons and
animations from .glb files.
Multiple animations are handled and will use the name in the GLB. The
default `viewer` process will end up playing back the first animation.
There are a few limitations:
- You can only have around 100 bones. It is technically possibly to have
slightly more, but certain animations may fail to compress when there
are more than ~100 bones.
- Currently, all animations have 60 keyframes per second. This is a
higher quality than what is normally used. If animation size becomes
problematic, we could make this customizable somehow.
- There is no support for the `align` bone.
---------
Co-authored-by: water111 <awaterford1111445@gmail.com>
This adds support for replacing existing merc models in FR3 files with
custom GLB model files. The replacements go in
`custom_assets/<GAME>/merc_replacements`, similar to texture
replacements. When a `.glb` file with a file name that matches any model
present in an FR3 is detected (e.g. `eichar-lod0` for Jak), all merc
model data is replaced with the given model.
Additionally, models for custom actors can now also be added to vanilla
FR3s. The models for this go in
`custom_assets/<GAME>/models/<LEVEL_NAME>` (e.g.
`custom_assets/jak1/models/jungleb/test-actor-lod0.glb`) and will be
added to the FR3 that has a matching name (exception: to add things to
the common level file, the folder should be named `common` instead of
`GAME`).
For custom levels, these now go in
`custom_assets/<GAME>/models/custom_levels` (previously
`custom_assets/<GAME>/models`).
Another small change: When level ripping is enabled, the resulting model
files will now be stored in game name subfolders inside of `glb_out`.
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](https://github.com/open-goal/jak-project/blob/f6688659f2ef85f5ceaacea6271580c9f4d91ed1/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 does a couple of things:
- The `custom_levels` folder was renamed to `custom_assets` and contains
`levels`, `models` and `texture_replacements` folders for Jak 1, 2 and 3
in order to keep everything regarding custom stuff in one place.
- With this, texture replacements now use separate folders for all games
- A build actor tool was added that generates art groups for custom
actors
- Custom levels can now specify what custom models from the `models`
folder they want to import, this will add them to the level's FR3.
- A `test-zone-obs.gc` file was added, containing a `test-actor` process
that uses a custom model as an example.
The build actor tool is still very WIP, the joints and the default
animation are hardcoded, but it allows for importing any GLB file as a
merc model.
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.
Patching up the extractor while working on the launcher, fixes:
- makes it so you can compile successfully given a folder path
(currently assumes your project path contains `iso_data`)
- ignore `buildinfo.json` from validation code.
- fixes an edge-case that could recursively fill up your entire
hard-drive!
- allows overriding the decompilation configuration via flag
- adds a way to specify where the ISO should be extracted to
- `fma-sphere`
- `prim-beam-h`
- `cam-start`
- `ragdoll`
- `light-trails-h`
- `light-trails`
- `menu`
- `water`
- `water-flow`
- `hud`
- `hud-classes`
- `progress`
- `progress-draw`
---
The `get-texture` macro replaces calls to `lookup-texture-by-id` and
`lookup-texture-by-id-fast`. The `defpart` macro detection was modified
to print a pair like `(texture-name tpage-name)` for the texture field
that gets turned into a `texture-id` constant. Only used in Jak 3 at the
moment, I'll probably go through the other games at a later point.
This updates `fmt` to the latest version and moves to just being a copy
of their repo to make updating easier (no editing their cmake / figuring
out which files to minimally include).
The motivation for this is now that we switched to C++ 20, there were a
ton of deprecated function usages that is going away in future compiler
versions. This gets rid of all those warnings.
Adds two new lump types: `cell-info` and `buzzer-info`.
`cell-info` takes a `game-task` enum value and defines an `eco-info`
lump for power cells.
`buzzer-info` takes a `game-task` enum value and an integer and
automatically calculates the required value for the buzzer task with the
formula `task + buzzer * (1 << 16)`.
When giving entities custom actor IDs in your level JSON, it is possible
to break entity lookups by actor ID if the actors are not sorted by ID
because `entity-by-aid` expects them to be in order.
This sorts the actor list by ID before generating the level file and
also checks for any duplicates.
This adds support for using enums in lumps using the new lump types
`enum-int32` and `enum-uint32`. Also adds these other new lump types:
- `water-height` (takes 3 meter floats, an enum and another optional
meter float)
- `eco-info` (takes an enum and an int)
- `vector3m` (3 meter floats + `w` set to 1.0)
This will create a folder like `decompiler_out/jak1/entities` and save a
JSON file per level with all the actors.
Also, it should hopefully make custom level building a little faster.
Some general improvements for the texture animator:
- Clouds are special cased, saving about 1 ms per frame
- Adjusting the amount of clouds now actually works.
- Fixed an issue with the brightness of clouds, and the way that they
fade out around the edges.
Tyler Wilding