DF-21 Forums

[Cal72521]

After alot of hours of testing different df rooms to find how and when homs' happen. I have discoverd it is a liner function not log or something.
Frist:
Take some graph paper let each square represent 16 df units, label width for y[up and down] next label distance/diamiter for x [side to side]. Finally take a ruler and find w/y=256, d/x=128. next point will be w/y=160, d/x=208; w/y=64, d/x=288 and so on. Draw a line between these points. This is the dividing line for homs. The event horizon if you will.
The Meat:
Homs are a function of wall width and the distance you view the wall from. The engine draws the entier wall you can view even if you can only see part, so hour glass shaped rooms will not let you cheat.
Also if one stands in the corner one can see more 'wall' at a greater distance than in the middle edge of a square room.
Remember continuous wall width is the key here, height and the amount of separate walls don't seem to make homs.
In practace:
Take the diamter of the room you want on the graph say 320. Find the width of wall,for a hom is ~24 acording to our graph, so 16 is plenty safe.
320df*3.14=1004.8/16df=62.8<63
Finally create a sector with 63 sides and a radius of 160.
or
A room 320df long and 16df wide that works to.
The largest room i've made so far is 320df units with 15.95 walls and 63 vertices. And DF it loads fine with 3dos and objects too. Even larger rooms could be made! A note at 288 and above, some clipping happens but no homs.
The largest practical size square room is 176x176.
homs= Hall of Mirrors
Diamiter= Viewing distance from a wall
Rooms= Sectors

Try it out!
You'll find homs happen at exact points.[/img]

_________________

[klasodeth]

Hmm... That sounds like it might be useful information for lucius. I thought I remember him saying that he planned some sort of heuristics for the DarkXL editor so that it could warn the user if a level might break Dark Forces compatibility. If so, knowing how to check for a HOM condition would be useful.

[Lionel Fouillen]

No time to test it myself, but thanks for the info!

[Barry Brien]

Good work must give that a try.

[Barry Brien]

Guys I've been trying this out on a new level that I've started messing around with, and to be honest I'm amazed. This works really well. Why didn't we figure this out earlier? I had a big area I as working on and I got to the point where it was homming on me when i went near any of the edges. So I divided up the walls and hey presto: hom is gone. So good work Cal, this is pretty important stuff as far as I'm concerned. Cheers!

[Burning Gundam]

You're working on something again? I hope you are. This could be good news.

_________________
I don't think outside the box... I customize it.

[Barry Brien]

Cheers Gundam, but you know how long I take to put stuff out so don't be holding your breath.

[Burning Gundam]

** Holds breath **

** Face turns a rainbow of colors **

** Dies **

_________________
I don't think outside the box... I customize it.

[Cal72521]

[Barry Brien]

I think that's probably the coolest fecking graph I've ever seen.

[Fenwar]

Fantastic work, and I can't believe that this was never discovered before (particularly having seen the level of research that went into the DF specs).

Anyone want to start a new HTML version of the DF Specs?

Or better still, stick it in the Dark Forces Wiki. Use it or lose it, kids.

[lucius]

It seems like this could be used by the DarkXL edit to split walls to avoid HOM (optional only - probably something you have to tell it to do). It could also flag walls that would be problematic. Good job.

_________________
DarkXL....http://darkxl.wordpress.com

[Cal72521]

[url=http://df-21.wikidot.com/hom]Dark Forces
Hom wiki made, its late so spelling is bad.
I'll research how many ajoins it takes to hom, or how ever that one works.

_________________

[Cal72521]

I did some tests with ajoins between sectors for homs, no subsectors. my graph holds true! i did sevral tests with walls 64, 128, and, 144. two tests at 64 where used, every 16dfu an adjoin, then every 8dfu. Doubleing the amout of adjoins. 32 then 16 adjoins no homing point corrilated to the above graph! euraka! if recreating these test it was found nessary to reskin the test wall. after hom point was exseded the next closest sector was drawn.
Its my opion that the gloden graph holds true for this kinda hom. there seems to be no max adjoin limt.
now subsectors are a diffrent story. I check that one next week.

_________________

[Cal72521]

ok did some testin!
you can NOT look thru two subectors in diffent sectors at once. like this O- | o | o | =hom.
also you can not it seems look thru more than 7 subsectors at once with in the same sector.
and of crouse you can't have a subsector within a subsector work.

Does anyone have any input? Thanks.

_________________

[lucius]

Those are interesting points to keep in mind. Those cases work just fine in DarkXL which means that the editor will have to eventually flag those cases with a warning if you want full DF compatibility. So the editor will have an DF compatibility option - which will warn in cases where it will HOM like you mention here. Of course you will be able to turn this off if you just want the level to work in DarkXL.

Thanks for all the testing.

_________________
DarkXL....http://darkxl.wordpress.com

[Fenwar]

Excellent stuff Cal, but maybe don't try to write when you're drunk?

[Cal72521]

oh df specs are so great to have.
there is a 40 adjoins viewable hard limt.
i didn't even know!

_________________