FS2004 Shadow models backgrounder

[Here for the Beer]

Hi folks, me again...

I'll soon be finally sitting down to make a shadow model for the F900 to remove the daft shadows (3 variations of aerials all at once along the #2 engine top). I wanted to learn a little background and looked at the threads here and elsewhere, the wiki, the FAQ etc and didn't really get my questions answered, so here it is.

Assuming I was even lazier than usual and made a shadow model that just omitted the aerials that offend me so much, would it

a/ make things worse by being a whole extra model at that LOD just to remove a few shadows of aerials, and/or

b/ need to be created for every LOD the final model uses, or

c/ can there be only one shadow model even for a LODed object?

What I'd like to do, to simplify the job, is use a separate shadow model only for the top LOD; the next LOD down loses the aerials that cast the daft shadow anyway. Once it is far enough away and simplified anyway, the shadow "problem" no longer exists.

So, is it simply a case of adding a suffix to the shadow model that makes it "appear" only at the top LOD, or does that approach mean that the next LOD down has no shadow at all.

I test my models across the course of the sim "day" and often see them at dawn and dusk. The shadows then are hard to simplify. If the gear is full of struts and doors and the shadow is simply a gear leg, or a door and a wheel, it wouldn't look so good, it would really jarr (at least, it'd have that effect on me). This means that there's a limit to what I'd want to remove (it is AI, the shadows are going to be easier to live with because the top LOD is moderately simple anyway). Pitots and dorsal aerials are the most likely to go from the shadow model because they're usually shadowing themselves on the plane, not the ground, until the sun gets very low on the sky.

I'm assuming at the moment that the shadow model doesn't take any frames for rendering the object, although the visible model is, and the visible model, if a shadow model is present, is relieved of the job of creating shadows altogether. Would that be a good assumption?

Cheers and TIA for any answers.

 

[Here for the Beer]

After some more experimentation, I still cannot get shadow models to even appear. I've tried making them per-LOD, for the whole model and even just a small planar object to see whether I could make the shadow disappear altogether.

I've read the MakeMDL instructions and followed what they say to no effect. There's also the possibility that the export process needs to be told there's a shadow model, but I can't find any reference to a flag that says "separate shadow model needs to be compiled", so presewntly the background information is less relevant anyway.

What I'm doing is creating a shadow model in the same way as I create a LOD. In fact I've experimented with cloning an entire LOD and just renaming it F900_shadow_100 instead of F900_LOD_100. Nope, doesn't do the trick.

En-route I have had the idea that I make a shadow model just for the offending aerials, which is what I plan on doing (they are only visible on the top LOD or two anyway), but the naming convention I'm using obviously isn't enough (even though that works fine for LODs).

Since simply renaming parts evidently isn't sufficient, does anyone have some advice as to what I'm either failing to see or do?

 

[Flusirainer]

You can in the asm file instruct which LOD as shadow will be displayed.


Scroll the contents of the file to the bottom!
LOD_11 is model with maximum detail

model_shadow label BGLCODE
    IFSIZEV    SHADOW_11, 3500,1000 ; lod = 12.000000
    IFSIZEV    SHADOW_10, 2500,1000 ; lod = 25.000000
    IFSIZEV    SHADOW_9, 1750,1000 ; lod = 50.000000
    IFSIZEV    SHADOW_8, 1200,1000 ; lod = 100.000000
    IFSIZEV    SHADOW_7, 800,1000 ; lod = 200.000000
    IFSIZEV    SHADOW_6, 600,1000 ; lod = 300.000000
    IFSIZEV    SHADOW_5, 450,1000 ; lod = 400.000000
    IFSIZEV    SHADOW_4, 350,1000 ; lod = 500.000000
    IFSIZEV    SHADOW_3, 275,1000 ; lod = 600.000000
    IFSIZEV    SHADOW_2, 200,1000 ; lod = 700.000000
    IFSIZEV    SHADOW_1, 150,1000 ; lod = 800.000000

SHADOW_0   label   BGLCODE
    BGL_JUMP_32    LOD_0L
SHADOW_1   label   BGLCODE
    BGL_JUMP_32    LOD_1L
SHADOW_2   label   BGLCODE
    BGL_JUMP_32    LOD_2L
SHADOW_3   label   BGLCODE
    BGL_JUMP_32    LOD_5L
SHADOW_4   label   BGLCODE
    BGL_JUMP_32    LOD_5L
SHADOW_5   label   BGLCODE
    BGL_JUMP_32    LOD_5L
SHADOW_6   label   BGLCODE
    BGL_JUMP_32    LOD_5L
SHADOW_7   label   BGLCODE
    BGL_JUMP_32    LOD_10L
SHADOW_8   label   BGLCODE
    BGL_JUMP_32    LOD_10L
SHADOW_9   label   BGLCODE
    BGL_JUMP_32    LOD_10L
SHADOW_10   label   BGLCODE
    BGL_JUMP_32    LOD_10L
SHADOW_11   label   BGLCODE
    BGL_JUMP_32    LOD_11L

LOD_0 has Shadow from the LOD_0 Model

LOD_1 has Shadow from the LOD_1 Model

LOD_2 has Shadow from the LOD_2 Model

LOD_3 till LOD_6 has Shadow from the LOD_5 Model

LOD_7 till LOD_10 has Shadow from the LOD_10 Model

LOD_11 has Shadow from the LOD_11 Model

These mappings can be freely configured.

 

[Here for the Beer]

Thanks Rainer - I was hoping that just naming the Gmax parts (as I do for LODs) would be sufficient, but if that's the only way, then when I get a bit more time to work on it in a single session I'll set up an ASM-editing workflow and try to get the models completed that way.

The MakeMDL instructions make it sound so simple!

 

[Flusirainer]

Small correction.

LOD_0 is model with maximum detail.

 

[N2056]

If it's just shadows from the aerials that bug you why not give them a specific material with the display shadows option turned off?

 

[Here for the Beer]

If it's just shadows from the aerials that bug you why not give them a specific material with the display shadows option turned off?

Ah ha! That might be the thing I'm looking for. The aircraft is AI so extra drawcalls are not normally welcomed, but in this case that might be what I can do - and it is only on the top LOD anyway. Since I have other aerials on the thing I could also use that extra drawcall to switch off shadow calculations for all those too - might as well make the best use of an adverse step and probably end up better off. I will try that asap - it also means I can keep the present process "as is".

I was looking at shadow stuff in the compilation phase - never thought to look at materials (the downside of exploring Gmax without a map and compass). Thanks for that advice, I appreciate it.

Edit: Except that my materials editor doesn't display advanced parameters. I hope it isn't gamepack-specific (I make stuff for FS9) and I just need to switch a mode somewhere (I tried expert mode and that just seemed to change the workspace). I've run out of time to check this for now. If I discover a reason later by Googling, I'll mention it here.

 

[Here for the Beer]

Well, I've spent most of today looking for this option and can't find it in my Gmax 1.2/FS2004 gamepack setup. I've found a cheap copy of the Gmax Bible online and might find something more in there. It is odd that nobody anywhere seems to be mentioning this "non-shadowing materials" option - I can't be that much of a trailblazer, surely!

 

[tgibson]

That option is FSX only, AFAIK. You can use MDLTweaker for removing shadows of FS2004 scenery models, but I don't know of one for aircraft models.

 

[Here for the Beer]

Thanks, Tom. When nobody on the interwebz is talking about something it either doesn't exist or is as boring as a particularly boring rock. Usually someone somewhere has had a problem before I have.

Of course, that has me wondering whether the feature would port back into FS9... I would guess not since a lot of FSX talkspace is taken up with shadows problems which indicates FSX tackles shadows differently (probably due to the overhaul of shadows with the advent of self-shadowing).

I will evidently take Rainer's hotwiring of the LODs as the best option.

 

[Flusirainer]

One possibility would be that antenna is provided with a material which is slightly transparent.
Transparent material produces no shadow in FS9.

 

[Here for the Beer]

Hi Rainer -

I will try that approach first to see whether I can get a sweet spot where it works without making the antennae get ghostly - at least then I don't change my workflow for this model and don't have to re-learn the process when/if I have to go back and revise the thing. The irony is that most of the antennae are hidden with alpha anyway, and it was the visual oddity of having an engine nacelle with possibly all the aerials hidden but all the variations still visible in the shadows cast by the aircraft.

I can actually live with it for my own purposes, but each model has to have a few learning experiences to make the creation process worthwhile - at least for me, to stop it feeling like a production line. Of course it gets frustrating for me sometimes but at least it is frustration I'm volunteering for! Anyway, I will try that idea - I considered the possibility a couple of weeks ago (not getting much time on the project at the moment) but was still committed to the shadow model idea using naming conventions in Gmax "_shadow" but not getting it to work. F900_LOD_100_shadow, F900_shadow_100 etc all seemed to do nothing - and I only need to knock off the aerial shadows on the top LOD since the next one down (85%) loses them altogether. Fun & games...

 

[Here for the Beer]

Finally got around to this - checking it out with all the 4 variants. May even be released this year (at the present rate of "progress").

Three aerial variations here - short spade. hook, and tall spade. There are actually 4.5 variations (the others are a hook with a small aerial-ette just behind it, and a "lump" aerial).

The shadows of the aerials are no longer on the ground thanks to making the material for them 90% opaque, but as you can see, they paint up just fine. Notice that the nose camera hasn't had "the treatment" and although it is masked out with alpha the shadow still appears. I've also fixed that and in passing removed the shadows for the pitots and other aerials. I assume that in the absence of a specific shadow model this is the de facto way to stop FS calculating shadows for small parts.

Well, works for me, at least...

Beginning to think about drafting that readme - always a good sign (depending on what you think of my models, at least!).

 

[gapeters]

Wow, that looks very nice. The number 2 engine inlet looks much better now.

Keep at it!!

Greg

 

[Here for the Beer]

Hi Greg!

I'm effectively done. There's some element of having to live with it in the sim in order to get a better feel for it, but also the need to get it released sooner rather than later, so I've done some last-minute things on the RV-8 while looking around the F900 in-sim. People that use previewers to look at their repaints will find this is the same as the F2000 in that the gear doors don't completely close (they get within inches and then disappear (wheelwell object in Gmax)). This looks okay in-sim, especially if you are actually aviating and not just following AI planes around in the sim.

Once the RV-8/A and the F900 are out the F2000 will get a minor revamp at the model level (existing repaints should be okay, the mapping isn't changing) to incorporate things I found when making the 900, and I will also make the F2000S variant, which for AI purposes is the addition of the F900's inboard slats to the F2000. I have made semi-sensible flightplans for them this time (bases are either accurate or at least feasible) and those plans will need updates since they were made about 3 months ago and a few F900s have changed hands since then.

Edit: Discovered a mapping problem on a lower LOD (tail area) that results in an obvious LOD shift - so there's some work still to do when I get the time. Presently major renovation upheavals here at home, so not a great time for Gmax development unless I can develop the ability to focus through the sound of power tools.

Edit 2: No workmen today (oiled floor drying) - fixed LODs.

Yay.

 

[gapeters]

Yay is right!!

Greg

 

[Here for the Beer]

Noticed another LOD texture glitch that required a similar fix, hence another delay. Just fixed it today and will check it out for a while in-sim, all 4 basic models needed it (900, 900A, 900W and 900WA - W=winglet, A=tail antenna) all have the optional aerial feature via alpha).

I wouldn't say it is perfect, I'm sure some people will find "features"! Still, it is near enough ready to go now. After this I hope to get time to fix the F2000 with the things I learned or spotted with the 900 (the 900 being based strongly on the 2000) and then move on to fix a couple of things on my light GA planes (some smoothing not done etc).

I have a few repaint/fp packages I'd like to put out too (including a worldwide Seneca package + some generic paints of my own) but time is narrowing down now and work is coming in. This years planned work is a mess thanks to the house remodelling - I keep forgetting that when the builders/fitters depart the scene, the work is only beginning for me!

Next summer I'll dispense with the bizjets and stick to light singles and twins.

Also, next summer I graduate, so the lead-up to that will be busy.

 

[Here for the Beer]

Ugh. I was transferring some parts back to the F2000 after they got a revamp for the F900 - and spotted an animation glitch with the 900 - one of the NLG rear door panels (the smaller one that is attached to the gear leg) stays in its bay on LOD 85 (too noticeable, can't ignore it). Amusingly the NLG was being ported back because I had fixed an F2000 animation glitch on the F900!

I know I have my work cut out for me, because it's me that cuts it out.

 

[Here for the Beer]

Readmes are being written, but things are stop-start because work is beginning for me again (final year of university). Also the Falcon 2000 v2 is being readied for simultaneous release (minor changes - "existing repaint friendly") so that process is a little more drawn out. Vans RV-8 will come out at the same time together with a few other minor projects (repaints and flightplans packages). Should all be within a month.

 

[gapeters]

Looking forward to all of these! Sorry it took so much longer than you had planned.

Greg