Floating-point comparisons are tricky beasts. See this article for a good summary of the issues involved:
https://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/
This PR improves support somewhat, although there are obviously situations in the code (such as comparisons to 0) that should still be addressed. This PR makes the following improvements:
1) Tightens the threshold on `IS_VEC_NULL` and `IS_VEC_NULL_SQ_SAFE` from 1e-16 and 1e-36 to FLT_EPSILON
2) Tightens the threshold for parsing `turn_time` in POFs
3) Restores the previous implementation of `fl_equal`
4) Adds a dedicated `IS_MOI_VEC_NULL` for moment of inertia values
5) Improves documentation
Follow-up to #4535 and #4443.
As part of the cleanup in preparation for model translation, there was a bit of code review that was actually incorrect. The calculation of the current step needs to be rounded down, because the step isn't actually complete until it fully reaches the next number. Rounding the step causes the submodel to jump around in all sorts of weird ways.
Follow-up to #4443.
* Add Animation Instance Flags
* Add SEXP barebones
* Unify and simplify animation parsing
* Clean up animation calling and use new animationlist api
* Add missing break
In #4443, `fl_near_zero` was set to use default epsilon values in the `IS_VEC_NULL` check. In this comparison, epsilon is the wrong threshold because it used to have a much smaller threshold. The fix restores the previous definitions of `IS_VEC_NULL` and `IS_VEC_NULL_SQ_SAFE` by exposing the epsilon value as an optional parameter (similar to how `IS_NEAR_ZERO` used to work. This PR also restores `IS_VEC_NULL_SQ_SAFE` where it was replaced by `IS_VEC_NULL.`
* more cleanup in preparation for model translation
1) Clean up stepped rotation, fixing some logic, adding safeguards for invalid values, and allowing rotations to go in the opposite direction
2) Add more specific SEXP error checking for AWACS and rotating subsystems
3) Assorted refactoring and polishing
* address feedback, plus a bit of extra cleanup
Fixed environment map rendering
shift fresnel term to match ordering of ComputeGGX, document shader code.
Envmaps: Use tangentMatrix to build local coordinate system
Fixed geometry self-shading calcs to improve shading of rough surfaces.
Cite paper correctly
optimsed env sampling
Put envmap face generation a loop to make it easier to read
Precalculate points on circle for even faster envmap sampling..
Guard against NaNs
Revert to simpler envmap calcs, significantly faster on slower GPUs.
typo
We don't need the sample points any more.
Diffuse IBL shading
Fix gamma issues in envmap rendering
AO term on envmaps.
fix lab crash with retail data
Always render envmaps - PBR shouldn't care if a shine/specmap exists.
Since "No subsystems found" and "Not all subsystems have a record in ships.tbl" are different warnings, they should have different flags. Follow-up to #4408.
The check to not show `-destroyed` submodels is within `model_replicate_submodel_instance`, and that was only called for moveable submodels in the lab. This PR runs it for all submodels now. Tested and fix works as expected.
Allow scripts to reference submodels, submodel instances, and model instances. Currently just a few properties (like Orientation and Radius) can be accessed, but this can be easily expanded in the future.
Also add relevant functions to calculate real-world positions and orientations of rotated submodels.
Consolidated these functions and replaced the `find_submodel_point_*` functions with their closest `world_find_` counterparts. (One of those functions may not have actually worked correctly.) Also use the correct coordinate logic when returning normals from collision info in scripting.
Related to #3985 although not yet a fix for it.
Mostly renaming of fields and functions, and moving some modelread sections into functions. This reorganization will put the code on a better footing for the model translation feature. No model translation code is included yet, and there should be no change in functionality.
There are a few related improvements here:
1) Add a new function for iterating over the submodel hierarchy
2) Add a counterpart to the `subobj_num` field so that a submodel can refer to its ship_info subsystem
3) Update `ship_model_replicate_submodels` to replicate all movable submodels, even the ones that aren't associated with subsystems
Fixes#3967
By anchoring a dockpoint to a submodel, the dockpoint will move when the submodel moves. Historically this has been done by inspecting the path associated with a dockpoint, but this method will not work if there is no path or no rotation mode specified. Here is an alternate method that allows the submodel to be specified explicitly through a dockpoint property.
Adam Leach