DirectQ and Ring of Shadows

[Magnus]

So not only is it high and to the right. Its high, right, and back.

[MH]

OK, found the problem.

Some history first. GLQuake doubled the size of the eyes model because it was harder to see than in software Quake. This was due to various limitations that GLQuake had compared to software Quake (in particular no overbrights), and is not really necessary in a modern engine, but it's what people are used to as being the "correct" size of the eyes by now, so we're stuck with it.

Anyway, this was done at runtime and controlled by the value of gl_doubleeys (note the final missing 'e') which defaulted to 1.

For previous versions of DirectQ I removed this for some reason (can't remember why) and moved the doubling to when the model was loaded. So far so good, it was consistent with GLQuake.

Recently I put it back (can't remember why I made that decision either) but got a few things wrong. Firstly I left the doubling at load-time in. Secondly, GLQuake applied an offset to the vertical position to put the eyes in more or less the same place they were in when not doubled. I accidentally applied it to all 3 positions.

Two mistakes which combined to give the result you noticed.

This is going to be fixed in the next version. For now just set gl_doubleeys (again note the final missing 'e') to 0 and it will be consistent with GLQuake/ProQuake/etc.

An interesting thing cropped up while investigating this. All GLQuake derived engines actually position the eyes slightly wrong when doubled; they're slightly offset in all 3 directions from where they should be. The correct translation factors are more like -1.41, -3.2 and -24. You can confirm this by quickly toggling gl_doubleeys on and off while looking at a player who has the ring. It's probably desirable to fix this if for no other reason than correct aiming.

[R00k]

TRue, the "double eyes" scales it, so it scales the bounding box too, center to the box is no longer center to the bounding box of the player model.

	if (clmodel->modhint == MOD_EYES && gl_doubleeyes.value)
	{
		glTranslatef (paliashdr->scale_origin[0], paliashdr->scale_origin[1], paliashdr->scale_origin[2] - (22 + 8));
		glScalef (paliashdr->scale[0] * 2, paliashdr->scale[1] * 2, paliashdr->scale[2] * 2);
	}

That +8 lowers them down a bit...

[MH]

It's not centered in the bounding box of the undoubled eyes model either, because all vertexes in MDL files are positive.

[Magnus]

Ok, the newest directq has the fix for the location of the eyes.

However, they are still SMALL AS ... you know what, heh.

In proquake they are HUGE in comparison.

I tried gl_doubleeys 0 and 1 and it looks identical.

At least this is a step in the right direction though.

Anyone know how I can make the eyes larger?

[Mindf!3ldzX]

stop using a 124256316363x321426458241 resolution. :F

[MH]

Anyone know how I can make the eyes larger?

You can't. It's a test release (that's why I called it "test release" ) and unfinished.

Breaking news is that the way GLQuake - and anything derived from it - handled the eyes is actually completely wrong. They're scaled correctly, but they also need an offset to re-center them at the correct position, and GLQuake does this in one dimension only and in a way that breaks with any replacement eyes model. This one needs more work.

[R00k]

Why cant one just rescale the actual eyes model in QME and call it a day....

Personally, I'd rather see a fully animated alpha-blended player model, at maybe 20% alpha..