Freeform Rework
No, not the full thing.
The familiar decision appeared again while reimplementing freeform with the new V0.2 editor, undo and input systems. Whether to bring it over while maintaining its current issues, or do another iteration on it.
Freeform was made about 4 months post release, when there was still a lot of internal pressure to demonstrate loyalty to the project.
It was rushed, it was met with a suitably unstable release, and now it has to be redone.
I've learnt from that mistake, which you can probably already tell from V0.2's current timeline.
Software development has to go slowly, otherwise it goes slowly.
I've since ripped out the old model generation code, which was an external dependency (built by other people). Replacing it with my own code.
Using a purpose built light-weight solution allows me to strip out everything that isn't needed for Sprocket. Resulting in performance boosts* and no hacky workarounds. Most importantly, it stops becoming a part of the code base I avoid, meaning freeform will get more love from now on.
* Only when editing compartments.
This is the first half of the freeform rework that's been discussed since the initial plan post for V0.2.
It implements all of the fixes, while leaving the new features for post V0.2.
- Inverted faces are a thing of the past. The invert button will remain so you can flip which side of the face thickness is added on to.
- Point and edge selection have been perfected. These should now behave as consistently as you'd expect.
- Part repositioning is now handled in a way which completely stops parts jumping around unexpectedly during certain combinations of operations.
- Undo redo has been perfected. Conversion from a generated design to freeform can now be perfectly undone. This allows the removal of the confirmation prompt and the 'Back to generated' action, as the undo button can be used instead. This perfection also applies to undo for part place and movement, where a few major issues spoiled its reliability.
Some tank-specialized model elements can now be thrown into the mix, which would be too performance intensive without the new specialized solution.
Plate Thickness
I did break the 'No extra freeform features' rule for this one feature.
It didn't seem right having the awkward gameplay phase where armour wouldn't consume space, as originally planned. Since people would churn out new designs with geometric internals, only to have them invalidated by the armour space change to follow.
Now interiors are nice and cozy.
Note this visualization doesn't mean there's weight balance concerns. Though that's a door opened by the track rework with the prospect of individually adjustable suspensions.
Automatically Generated Rivets
Another potential unlocked. This wont be done in 0.2, but it's soon to follow.
Expanding the Team
The plan is to expand the development team after 0.2.
In order to make that actually result in increased productivity, I'm taking special care to tidy up the code base and making some shifts to allow multiple programmers to work on it simultaneously with fewer issues.
There's a certain artist hard at work on the models to replace the placeholders of the internals - so there's some more exciting progress shots to come!
Hamish