- Messages
- 349
- Country

Not writing about a problem, just airing a kind of bloggish account of the past two days' work.
A while back I was fretting about a canopy transparency that was hiding the flaps on my AI aircraft, and discovered that grouping some completely different parts of my project made the masking problem disappear.
Well, as I suspected it would, the masking problem reappeared for some equally obscure reason and I set about trying to make a fix that would stick. This is where the fun started. I read about zbias being used for texture names and figured I would try it with part names.
So I called l_flap "zbias_01_l_flap" instead. I lost the free animation (oooh thanks, I'll have two) but the problem vanished. I then realised the problem was also visible on the fuselage beneath the canopy (I could see forward into the "engine bay") so I started fiddling with zbias on the fuselage. Fixed that problem but then the crew never appeared because part of the canopy is attached to the fuselage and was now taking priority over my disappearing-and-animating crew. Spent a night pondering a workaround that didn't mean compromising my animations etc and restarted work this morning. A few more experiments and nothing was changing for the better.
So, I stripped back the project, reverted the fuselage and canopy to an earlier version and started again. After a while I had something that seemed vaguely okay but the crew were still AWOL although everything else was still there, with no weird fuselage holes and flaps that were not animated but at least visible.
Started fiddling with the canopy to try and get the crew to appear, and finally ended up with the inner and outer layers of the canopy attached to the fuselage and the crew appearing. The downside was that I now had a multimaterial that I really didn't want to see in an AI project.
So, I thought to myself, "there's nothing to lose, I have plenty of fallback saves that I've made en-route, I can always revert", so I assigned the combined fuselage and canopy to the standard material for the one BMP the project uses. The multi-material had been created because the canopy uses transparency and, um, the fuselage doesn't, so it was effectively 2 materials for one object, except that now I no longer had material transparency for the canopy, just the alpha channel. I thought I was going out on a limb with this, but for some strange reason it actually worked. What's more, the fuselage and flaps had no vanishing sections.
Being shortcut-guy I thought "what the heck, I'll revert the flap names to lose the zbias_01_ prefix before I start in on creating a dummy object to get the wrongly-named flaps to animate again. Just to see what happens."
Well there's a thing. I have normally named flaps, visible crew, no fuselage holes and I'm back to having no zbias_01_ prefixed parts at all! I'm back where I started except that the canopy no longer has transparency specified in the material (so the project is actually reduced by one material) and both inner and outer parts are attached to the fuselage when previously only the outer one was attached - because the inner one hid the crew last time it was set up this way!
I'd swear that if the wind was blowing easterly one day, I'd get a problem after compilation, and when it switched around the darn thing would compile just fine.
My model is now simpler and works, it seems I just had to run around a loop of experiments to get back to where I was before, except that now it works. I can't help wondering whether this "fix" will be as temporary as the "group a few bits and see what happens" approach.
I think I'd better do all the LODs and versions in this Gmax session (perhaps leave it active all night), because the act of shutting down Gmax is probably enough to upset the whole house of cards.
And before anyone mentions it, yes, I have read the part in the manual about exporting models during the full moon.
A while back I was fretting about a canopy transparency that was hiding the flaps on my AI aircraft, and discovered that grouping some completely different parts of my project made the masking problem disappear.
Well, as I suspected it would, the masking problem reappeared for some equally obscure reason and I set about trying to make a fix that would stick. This is where the fun started. I read about zbias being used for texture names and figured I would try it with part names.
So I called l_flap "zbias_01_l_flap" instead. I lost the free animation (oooh thanks, I'll have two) but the problem vanished. I then realised the problem was also visible on the fuselage beneath the canopy (I could see forward into the "engine bay") so I started fiddling with zbias on the fuselage. Fixed that problem but then the crew never appeared because part of the canopy is attached to the fuselage and was now taking priority over my disappearing-and-animating crew. Spent a night pondering a workaround that didn't mean compromising my animations etc and restarted work this morning. A few more experiments and nothing was changing for the better.
So, I stripped back the project, reverted the fuselage and canopy to an earlier version and started again. After a while I had something that seemed vaguely okay but the crew were still AWOL although everything else was still there, with no weird fuselage holes and flaps that were not animated but at least visible.
Started fiddling with the canopy to try and get the crew to appear, and finally ended up with the inner and outer layers of the canopy attached to the fuselage and the crew appearing. The downside was that I now had a multimaterial that I really didn't want to see in an AI project.
So, I thought to myself, "there's nothing to lose, I have plenty of fallback saves that I've made en-route, I can always revert", so I assigned the combined fuselage and canopy to the standard material for the one BMP the project uses. The multi-material had been created because the canopy uses transparency and, um, the fuselage doesn't, so it was effectively 2 materials for one object, except that now I no longer had material transparency for the canopy, just the alpha channel. I thought I was going out on a limb with this, but for some strange reason it actually worked. What's more, the fuselage and flaps had no vanishing sections.
Being shortcut-guy I thought "what the heck, I'll revert the flap names to lose the zbias_01_ prefix before I start in on creating a dummy object to get the wrongly-named flaps to animate again. Just to see what happens."
Well there's a thing. I have normally named flaps, visible crew, no fuselage holes and I'm back to having no zbias_01_ prefixed parts at all! I'm back where I started except that the canopy no longer has transparency specified in the material (so the project is actually reduced by one material) and both inner and outer parts are attached to the fuselage when previously only the outer one was attached - because the inner one hid the crew last time it was set up this way!
I'd swear that if the wind was blowing easterly one day, I'd get a problem after compilation, and when it switched around the darn thing would compile just fine.
My model is now simpler and works, it seems I just had to run around a loop of experiments to get back to where I was before, except that now it works. I can't help wondering whether this "fix" will be as temporary as the "group a few bits and see what happens" approach.
I think I'd better do all the LODs and versions in this Gmax session (perhaps leave it active all night), because the act of shutting down Gmax is probably enough to upset the whole house of cards.
And before anyone mentions it, yes, I have read the part in the manual about exporting models during the full moon.

