P3D - a framework for 3D applications

Quote:

---

Originally Posted by schrodinger

@ happygreenfrog:

Ohhhh I did miss your latest update to Q.A. 3D, way cooool!
I particularly liked the dizzying part, great idea :D



@Lakera:

you got the point - no skyrim!
But as you say, imagination and artistry could led to interesting new gaming "perspectives" X)



@ jamesnansen (thank you for the idea!)
and everyone interested

we're going to have dynamic light sources:

https://www.youtube.com/watch?v=IprZR7bWlMw


this is a first draft, made with additional objects
(but I'm probably going to use a qualifier for this: "glowing" will be perfect)
we'll have to use some extra values on static sprites and physical objects for "light_radius" and "light_intensity"

with 4-5 lightsources it's dragging some 3-4 extra fps
but still needs to be properly optimized
I'd say that's bearable for the engine
will surely do for next build :)


(you may have noticed the light passing through walls - avoiding this would be way more intensive, I fear)

EDIT:

forgot to say (and show) you can also dynamically alter radius and intensity:

http://lizardking.co/P3D/P3D_torch.gif

---

Wow, that's neat! :D

Well I had to just buy to congratulate on such an impressive piece of work! I still finding out things about with this, (although I really shouldn't as I need to finish my work!)

I find some of the platforming to be a little buggy. Pressing right while against a "block" makes you push backwards. Also some of the animations refuse to work sometimes. (Plus weirdly, it's lags/stutters until you've seen all the level, then it runs at 60.)

but other than that it's pretty nifty!

Thought I'd upload a little image of what I've managed to create with it ;)
http://orig15.deviantart.net/ce24/f/...ge-d9muaqe.gif

Inspiring Sinclairian,

You made that with P3D!!! Cool!

Paper Mario anyone?

Quote:

---

I created a "test" mfa file of the P3D as you suggested. I'm trying to create some different ground static sprites, like roads for example. I copied an image of a road, I cloned the "grass" object, edited the "grass" object to become my road. The road's size is set to 64 x 64 with the animation direction down like suggested. The road has all the qualifications and settings as the "grass" object but when I put it in my frame and test it, I don't see the road. Now, I see that the "grass" object has a "3d_grass" object and I'm sure this has everything to do with making the "grass" object 3d viewable, maybe?

---

Hey Emerson,
I hope you're having fun with the engine :)

...well, beside this little annoyance! X)

3D objects are made with 2 active objects:
one being the "map" tile (qualifier= group. 3D)
the other being the "texture bank" (qualifier= group. data)

From what you say, I would make sure that:
1) the "map" element has the correct "texture bank" name referenced in my_model string
(must be the object name of texture bank, the one you can set by right-clicking- "rename" the object in frame editor)
2) the hot spot for your road in the "texture bank" element is placed in the bottom left corner

if you already checked this, or this doesn't work,
feel free to send me your .mfa --> support@lizardking.co

I'll check out and see what may have gone wrong!

Quote:

---

Originally Posted by Sinclairian

Well I had to just buy to congratulate on such an impressive piece of work! I still finding out things about with this, (although I really shouldn't as I need to finish my work!)

I find some of the platforming to be a little buggy. Pressing right while against a "block" makes you push backwards. Also some of the animations refuse to work sometimes. (Plus weirdly, it's lags/stutters until you've seen all the level, then it runs at 60.)

but other than that it's pretty nifty!

Thought I'd upload a little image of what I've managed to create with it ;)

---

Thank you Sinclairian!

You're absolutely right about the issues and thank you for spotting those out!
I'm very active on this project and it's still in its 1.0 stage -
so it's likely to grow/be fixed in the future basing on your valuable suggestions guys ;D

I'll be working on some animation system soon, btw,
I was thinking dedicate a value for that but wasn't sure if this would have been less intuitive
(but prbably more solid)
we'll see!

Aaand..... wow, it's beautiful!
Amazing to see how talented graphic designers may use your tool, I feel moved

"the "map" element has the correct "texture bank" name referenced in my_model string
(must be the object name of texture bank, the one you can set by right-clicking- "rename" the object in frame editor)."

This is the problem!

Dude, this was probably the best Christmas present I received. Can't wait to create!

Quote:

---

Originally Posted by schrodinger

Thank you Sinclairian!

You're absolutely right about the issues and thank you for spotting those out!
I'm very active on this project and it's still in its 1.0 stage -
so it's likely to grow/be fixed in the future basing on your valuable suggestions guys ;D

I'll be working on some animation system soon, btw,
I was thinking dedicate a value for that but wasn't sure if this would have been less intuitive
(but prbably more solid)
we'll see!

Aaand..... wow, it's beautiful!
Amazing to see how talented graphic designers may use your tool, I feel moved

---

It's very amazing little thing, I've been having fun stress testing it! It's great for making those old retro "first attempt at 3D" games much like Starfox and Virtual Racing. I would love to make a 2D platformer but with-in a 3D world, or perhaps a FPS Rogue Like with this little tool, however once it's a little more stable.

Also I do have a question, is there a way to disable directions in the 8-directional movement? So it's just left and right?

Keep it up, I can't wait for an update! Even though I'm now transferring to an actual 3D development software, this is much much more fun and amazing!
-Joel

Thanks :) I'll keep up improving this!

Quote:

---

Also I do have a question, is there a way to disable directions in the 8-directional movement? So it's just left and right?

---

Not in a "elegant-builtin" way,
but very roughly, you can locate the "custom movements" group
and inactivate lines:

595 & 596

599 to--> 602

should do the trick!

(if you have moved something and line numbers doesn't match, it's the move up/down etc. events under the "8 directional movement" subgroup of "custom movements")

Just a couple quick news:

Quote:

---

Originally Posted by Zonacas

schrodinger

Would it be possible to make an top view as in the GTA 2 game?

---

Zonacas,

sorry it took me a while because I didn't have fitting textures nor sprites available
here's an example of something along that line:

https://www.youtube.com/watch?v=V_VaZDg27Qg

(actually, for top view, I should have better used another "top-view" car-sprite, but just a quick test..)


Aaaand, very good news for P3D users,
seems like Butterfingers found a great solution to avoid fading transparency
(and open to new possible "fogging" colors!)

http://community.clickteam.com/threads/95658-Doom-Wolfenstein-Style-Game-%28made-with-P3D%29?p=682999&viewfull=1#post682999


Thanks for your interest guys!

STAHP IT.

You're giving me more reason to buy it :P. I'm chomping at the bits, but have to wait a few paychecks :P.

Wow! Looks like you could make GTA 3 as well! XD

Schrodinger i'll set up a paypal account ASAP and buy this so I can give some feedback (who doesn't have paypal in the current age XD). Congrats on the actual release, btw :)

Thank you all guys,
I hope pricing is acceptable,
and hope P3D range of use/limitations being clear enough,
with specs and examples video published online
(gonna add more -and tutorials-!)
I love doing things for free, but this was 8 months of work or so...
and still working XD

There are things to improve, things to fix and new routes to try,
users interest makes me twice as determined in perfectioning this tool.
And most of all I'm happy hearing it's fun to play with X)

So - just wanted to say (again) thanks everyone for your support & feedback!

This guy is genius, trust me! :)

Amazing job schrodinger!

Honestly shrodinger, I see this as a new milestone for CF, I haven't seen anything like this before.

Guys, you're too kind, I'm flattered :O
and committed to improve this tool

Also wish to thank MuddyMole every 2 thread-pages or so XD for doing Quad shader,
it was a key step for making P3D come true.

I so want to pull the trigger on this but I have to wait. I've been looking for something that could be used to make a Doom RPG styled game, but I need to wait a bit. Seriously impressive work with this, it's staggering.

This is amazing work schrodinger. I have a free weekend ahead of me, so I plan on spending it playing around with P3D.

Bought this as soon as I found out about it yesterday, will be giving it a thorough go-over within the next few days, good work man. And about your pricing, good show on that, given what you've made here I'd say it's more than reasonable.

Warmest Regards

this looks so cool - I'll definitely have to buy it to play around with - great work!

very cool engine, I have a question though that isn't answered in the manual. What is the value that I need to input into a slope object to change the slope angle? looking at the code I can see its in P3D_slope_calc but I'm unsure if its Slope_Length or what. It'd be a really nice feature if that value could be adjusted from the object itself, as it'd be pretty damn amazing to create some kind of Landscape sculpting program (you already can make this, but a bit more resolution in the slopes and it could look really something with no expense to resources.. just means you have to be smarter with how you design your levels).

Can't believe how much power this has, all without an extension.. Your documentation is great, but I really wish you had included a variable reference sheet so we know what each Variable does.

EDIT: Also I discovered that the engines performance has no impact by screen resolution. You can make the frame as big as you want it has no impact on performance. The impact seems to be coming from the Textures.

Thank you Shabadage, Starvingindie, James555 and DaveC!
Hope you'll enjoy the engine! Let me know for any doubt or question before (or after!) purchase.
Also, new version will fix some bugs and add new features soon.

MrCyberpunk: I'll be adding more infos to the manual as well!

For slopes, if you are using the ones included in the sample frame,
they are all "custom" objects and you need to change the "3D Data" object Z coordinates offset to change the slope percentage.
This is simpler as it may seem: in the "3D_slope" object you will find two "offset" coordinates, should be _Az and _Bz.
They are set to 32, this means those coordinates are "pulled up" by 32 pixels.
Simply rise this number to increment the slope percentage.

Another way, not documented in the manual, is simply creating a "flat" 3D object, like a grass tile or whatever,
setting obstacle_type to "slope"
setting action point at the bottom, and setting its X angle to any value: this will become the slope percentage
(in short, you are rotating it through its X angle to make it inclined)

Only drawback is you will need to calculate its ending position for adjacent pieces,
while with the custom shapes included it's easier and probably cleaner.


OH - The variable reference sheet should be in the end of the manual, the "blue-white striped" tables, is this what you were after?
Though I noticed "preset named" variables are missing, I'll be adding some more info in there too!

Actually, I've tried to stock most needful info in the manual but there's much going on behind the shell,
so it's been hard keeping track of everything X)
Will try my best to make it cleaner and more comprehensive in next upgrades!

ah yes you have done a reference sheet, awesome. It is however missing the bits you just talked about. But that's great info. Really interested to see what kind of things I can do with this.

Me too!
Would be happy seeing things you talented guys will do with this tool :)

And thanks for your feedback - I'm very interested in knowing what's missing or what can be better explained in the manual, so I can improve it.

I just bought it, sorry for the n00b question, haven't spent much time with it... but if I have a 3D cube, then I manually set the size, say, to 25W 25H - the collisions no longer work correctly, is there a way to redefine the collision width/height etc?

Bloody hell this engines powerful. If you don't use Polygons and rely entirely on the 3d sprite rendering capabilities, you can get literally several thousand sprites working with minimal lag, and resampling on. Immediately what comes to mind is something like Wing Commander as that was entirely sprite based. This is much better than the Raycaster as the ray caster didn't allow for Z positions, nor freelook.

Quote:

---

Originally Posted by DaveC

I just bought it, sorry for the n00b question, haven't spent much time with it... but if I have a 3D cube, then I manually set the size, say, to 25W 25H - the collisions no longer work correctly, is there a way to redefine the collision width/height etc?

---

Hi Dave!
Sorry for the delay,
I assume you set W and H in the "map" 3d object dimension (or _W_/_H_ variables?)?
3D objects "real" width and height are directly withdrawn from the "data" object (the one referenced in "my_model" string)
but collisions are detected on the "map"
so you have to draw both your "data object" cube sides as 25x25, and your "map" object dimensions as 25x25

Hope this is clear (also check the "creating 3D objects" section in the manual for this)

This can be little intuitive at start, and is something I would like to cover in a quick video tutorial on the engine basics soon

Quote:

---

Bloody hell this engines powerful. If you don't use Polygons and rely entirely on the 3d sprite rendering capabilities, you can get literally several thousand sprites working with minimal lag, and resampling on.

---

I spent lot of time optimizing stuff, but polygons are easily the heaviest thing to process
(need 6-10 times the calculations needed for sprites + the shader application is really intensive,
pulling out the shader - resulting in unwatchable blocky polygons XD - gives a powerful boost).

Also, only things falling in the view area are processed, everything else is "switched off" and causes minimal bulk
(this is why "render depth" is so important to boost performances)

Btw, noticed a little bug on static sprites shadows not being switched off when their parent is off screen, and hanging around the layer..
fixed in next version >:)

I loved first wing comander, spent hours fighting Kilrathi ships on a crappy 286

I'm not sure about this... parting with money I mean. But this looks awesome! (love Sinclairian's platform example on page 5.) Will definitely buy it soon. Just don't put the price up :)

Caesar or anyone.

I promise it's worth it. It adds, not meaning pun, a new dimension to Clickteam. Usually I buy a bag of chips or a snickers once a day per week. I gladly sacrificed that week for P3D. XD. Big sacrifice. Lol

You mentioned before texture correction could be added for a Wolfenstein 3D-type game. Would this also be the case for a mode7-style game, meaning it would be one big tile-based plane?

eheheh price is fixed ;D no worries
(...unless huge additions/rewritings in the very far future...
but they could probably do for another different release though)


@ Emerson: I'm happy saving you from a week of chips & snickers! XDXD
Thank you for your kind words!

@ Hayo: unfortunately not.

The Wolfenstein advantage comes from having straight vertical walls and no X-rotations,
to have the same advantage on ground textures,
you should only have them "parallel" to the screen, and no Z-rotations (turn left-right)
-much like this old NES title we were talking about yesterday-
in this situation, with a modified version of the shader MuddyMole already prepared, we could apply a clean clipping.
But you see - this is a very limited situation.
For Wolfenstain it's good, as sacrificing ground and ceiling anyway makes for a good interactive "maze".

I've spent a lot of time testing new systems and tweaking P3D's system: "backprojection"
(ask Happygreenfrog on the matter X))
to allow for bigger textures and cleaner near-distortions.
I don't give up and will try testing new ideas, but as of now can't see a big improvement on this side coming soon.

I bought it earlier. You can put the price up now. It'll still be worth it :)

Great work Schrodinger. Not done much with it yet, but the examples and PDF are great. Superb so far :)

Wow, thanks Caesar,
you guys are making me proud of my little child X)

I'm very happy users like it and still more, have fun with it,
this boosts my determination to improve P3D as much as I can :D

This is impressive. Good job! I usually use Unity for 3D games, but I bought this just to see how it works. Awesome that you managed to get it working with no extensions and just one shader.

I'm sure you're aware of the texture distortion --- this is because the shader is rendering the quad as if it's just a 2D shape on the screen. In reality, you want your pixels to get more dense with distance.

http://i19.photobucket.com/albums/b1...correction.png

You could write another shader that applies this correction, but it would be more complex and possibly slower. Still, it's an option you could think of adding in the future.

Walrus - thank you for checking P3D! :)
I've seen your awesome work on your website and was curious about your thoughts, as an experienced 3D designer.

Yes I'm aware of the affine, non perspective distortion of the shader (this is entry number 2 of "Limitations" chapter XD)
Unfortunately, I've researched a lot, and consulted with 2-3 great shader experts,
and it seems that true perspective deformation with a pixel shader is not doable
(on the contrary, it's easily doable with vertex shaders)

..and even if some recursive processing to split the texture in affine sub-units,
(making for more perspectively-consistent rendering)
could possibly be tested,
with the limited power at our disposal
and with the shader "version" support Fusion currently has,
this would have been forcedly a very small/limited process,
one could better do "manually" by preparing smaller textures (32x32, 16x16...)

If you have different infos on pixel shaders perspective distortion capabilities, I would be very happy knowing!

I've written many shaders and I think it should be possible. The shader would need to be designed differently than it is now; it most likely needs to have more knowledge of what's going on in 3D.

I'd have to work it out on paper to visualize the actually correct way of doing this... but look at it this way. For any quad that isn't totally perpendicular to the camera view (so it'd be invisible), there exists a function that, using perspective-correct 3D matrix math, transforms a UV coordinate on the quad to an XY coordinate on the screen. Since the mapping is one-to-one (i.e. no two UV coordinates give the same XY coordinate), I see no reason why you can't figure out what the equation is, then use algebra to reverse it, so the XY coordinate of the pixel you're rendering in the shader gets turned into the UV coordinate you sample from.

I have a very limited experience with shaders (Quad was skillfully crafted by MuddyMole),
so I can't tell in appreciable detail why similar approaches ended up without luck.

I remember other tests feeding an additional "w" (depth) parameter
and getting nice but "unconsistent" results: taken singularly the quad was depth-scaling appreciably,
but in the whole context the overall result looked "wrong" near other quads.

Beside this past experience,
and the higher number of calculations needed, possibly making for a not bearable / very slow result,
I can't tell whatever else could go wrong.

If you think you could give it a try,
visualizing and making some sketch to see whether it could be doable or not,
I would of course be very interested,
and if you think it's doable, I'm available for a (reasonable X)) reward for your valuable services.

So, if you would like/have time testing your ideas (now or in the future), let's get in touch!
This would surely solve some current limitations too
(like the "homogeneous" texture sizes requested for an appreciable scene rendering)