Spryke - hardcore new-school platformer

Ok made up a simple example with some platforms more in "the Spryke way"
(well, if you use a good amount of imagination to see awesome graphics when there's now crappy sprites XD)
and some random 3D objects placed here and there:

Please login to see this link.

Warning: default Fusion platform movement, getting stuck is behind the corner XD

preview gif:
Please login to see this picture.

A very plain level, with some random ideas like a "trap" platform rotating down towards the end
(I think it's a nice fx because you don't have to animate it, and so it's flexible to behave the way you want)

A random "platforms field" of 100 elements is created towards the far right of the level, to test performance in crowded environments.

There's some displacement maps, I found them damaging performances with "resample" ticked, but maybe it's only my machine (see performance below)

The way it is now all the "3D" stuff is behind the top layer, where the player and 2D platforms are.
With some little customization, some 3D stuff could be put even "above" the player/platforms.

I've left P3D simple "debug" box open on topright of screen so you can enable and disable "resample" from there,
and also change render depth (how far you can see) with +/- buttons.

Only the tiles where I've written "2D tile" are the real flat 2D obstacle platforms,
against which the 2D hitbox collides as it would do in any plain 2D game.
You can see the trick by using Q/E and/or A/D keys, those will tilt camera X and Y angles,
so the 2D platforms will go out of sync with the 3D world, to resync again press "enter".

On the editing process:
I won't say this is exactly that comfortable to do on a large scale with plenty of platforms XD
but don't know either how much hassle is making those "pseudo-isometric" platforms by hand...
spent a couple hours making up the short example you see, though I can probably improve some aspects of the process.
So basically I think it's worth considering not as a "time-saver" solution, but only if the visuals are pleasing enough
(on that side, I guess the shaders could be slightly modified to improve on the antialiasing side, but should check better in case it's needed!)

On the performance side:
the file is capped at 200 fps, on my not that good pc it runs at solid 200 fps when not scrolling, at 180 when scrolling
worse performances with "resample" ticked when the "displaced volumes" are in sight, going to about 130 fps
and when there's about 100 platforms in sight (far right on the example), going down to 150 fps.
When going full screen (alt+enter), it's almost always at full 200fps, but our good old microstutter is more noticeable here..

There are some other things to consider, possible issues with the depth buffer and resolution changes etc. (you could probably avoid the depth buffer though)
but won't go too much in detail as you might not like the result anyway.

If on the other hand you like the idea, feel free to ask more info,
I've spent some considerable time working on this kind of stuff and I'm happy to help Spryke if I can,
even if you think you want to do it and end up not using P3D for that ;D

[MENTION=18866]schrodinger[/MENTION] - does P3D feature camera culling ie. blocks off screen are not rendered? You would likely need to render 5-20 platforms if so...

@Please login to see this link. : thank you for supporting P3D!

I definitely understand how implementing this solution could be a delicate decision,
and sure if you like it you're good giving some days to an in depth try to see what kind of problem could arise.

The example I've posted here is made with 2.1 version, which is about to be published
(I'm saying this since a couple weeks but now I promise it's true: I ONLY have to tweak minor things and prepare some other example and update the manual XD
I spent a couple weeks more to add triangles though, so I think it was definitely worth the wait!)
Despite the fact this time I'm just adding a +0.1, the new version dramatically improves the engine in a number of ways,
including 99% of the things you see rendered in that example
(previous versons didn't have box,sphere,cylinder,displaced volumes and triangle primitives,
all was only made with quad meshes and some limited "isosceles" triangle)
and also features among other things this simplified 2D-3D sync feature.

I can send you the engine beta mounted on that example if you want to tinker with it before release,
it's pratically ready, maybe seeing if I can streamline the 2D-3D process a bit more,
but that example was put up quickly and is VERY messy!

Unfortunately I'm about to leave now but will be back later with detailed answers to the questions.

Btw, I've recently open up a discord server for P3D support X),
not publicly spreaded the link so far but I was planning to do it shortly and include it in the manual for 2.1 update..
so if you prefer having a talk and we catch in a good timeframe you can join here:
Please login to see this link.
(other P3D users or interested wanting to join feel free too, of course!)

@Please login to see this link. : yes P3D does (try to) render only what's currently visible (if you look at the numbers on top of the screen, where you can see i.e. 50/125, the left number is the poly in sight -actually single primitives-, is it a singular mesh, not in the example save for the big water at bottom, or single primitive shape like the cubes).
I said "try to" because it's a delicate balance between what's faster: computing if a full body comprised of a number of segments and vertices is within an angled view frustum,
vs calculating if a cubic bounding volume is inside another cubic volume. So far I've found that the shader is much faster to process some thousands pixel more than the runtime calculating some hundreds vertices to discard the object! XD
So it does it, but in certain situations and with certain shapes, it does a bit coarse to save on calculations.

---I need to see this in a 1920x1080 environment. Partly to see the effect on performance, and partly to see how the 3D looks with more pixels. I wonder if the perspective will get more extreme with more horizontal resolution.

I honestly never tried P3D on very high resultions (primarily due to my monitor not supporting it XD) so that's something that would have to be tried, sure left-right edges of the screen will get a bit more squeezing but it should hold fine I think. Easy to try and see in case!

---Is it possible to adjust how much the sides of the platform tilt? Perhaps with a POV adjustment or something like that? I can imagine I might want a more 'subtle' effect that doesn't necessarily use a realistic perspective, but just tilts the platforms slightly.

I think there are some factors determining the "amount" and "look" of the effect, all of these you can customize, mainly:
- the "zoom factor" (how much objects are magnified)
- the platform dimensions >> depth, in the example you can see some are "thicker" and go more in depth than others, of course deeper the platform more noticeable the effect
- the camera position, to have a good 2D effect it must be perpendicular to the player and the "straight line" he moves on, but you can choose to offset the camera up-down/left-right and this will basically change the POV and so have an effect on how lines will go towards the vanishing point in respect to the screen

---One of the biggest questions is how feasible this is with the 'organic' platform shapes I intend to use. Below is a shortlist of shapes I'm currently considering. I'd say that this is pretty close to the type and variety of non-square shapes the final game will have
Is this likely to end up being prohibitively complicated, trying to get these shapes to work? If I paid you, would you be able to write code to make these shapes work, the way you've done for spheres, or is there simply a mathematics and/or performance barrier that puts that out of the question?

Looking at the shapes, many of them seem to have very low curvature, I think that could be feasible with meshes, being quad or triangles, probably quads would be good enough if the curvature is only on one, or opposite sides, like all of those thumbs I think? To get a smooth curve you need more polygons, so depending on "how smooth" the effect needs to be this could be somewhere in the 20-30 meshes per object (curved on both sides), doable, but could be problematic if lots of such elements are in the screen at the same time (for performances) and such pieces of course are not too handy to code ("model") and texture properly (unless they are all made from the same "material").
Pure speculation: I *think* some sort of parametric curve shader could be made, but it would be probably lot of work, and not so easy to use on already existing shapes (you'd need to somewhat obtain the curve equation form the picture...) so I would keep this option out to be safe now.

--by the way, there's a bug when you jump off the uppermost platform (the one that almost comes on-screen at the end of your gif) - the platform face moves with you

Good catch! That's the scrolling screen area hitting the top of the frame area and stopping,
it can be easily solved having a big enough frame area and moving all the level platforms down
(since you can't grow up the top of a frame area, nor "virtual scroll" to negative Y)
so it's something to plan at start: make sure there's enough room to scroll on the edges of the level

--Is there a way to add AA (or stronger AA, if there already is some AA) to the edges?

This is something I have to think upon, I see what you mean, edges are "pixel-crisp" and not smoothened (P3D2.x is a "raytracer" - the ray hits the shape or not, so there's no smooth "inbetween"). I'm thinking I could leave a "border" area to smoothen the color a bit (maybe with some alpha?) but can't properly answer until I make some test, I never digged this specific problem so I might need to do some additional research.

--I also don't have any understanding of how the editing process would work, and how well it would fit in to my current engine. I guess I can only learn that by playing around with P3D myself, which I'll try and do today.

Don't know if you tried, there was no easy way in 2.0 to make this trick (unless you made some customization to the engine)
but in 2.1 you'll be able to center camera on the player (or, say, your Volcamera X,Y positions) in any 2D layer, then passing the X,Y coordinates to the engine.
Setting up the scene consists in having a camera in the map looking straight north (Z_angle=90), and linking X,Z coordinates of this 3D camera to X,Y coordinates of the "flat" scrolling 2D camera.
Depending on zoom factor, the Y position of the 3D camera must be calculated for it to be perfectly in sync (so that 1 pixel in the 2D layer == 1 pixel in the 3D layer), it's just a one time calculatin, set up at start and leave it.
Once all this is well set, placing 3D platforms would follow the same placement you have in the 2D layer for X coordinates, Y coordinate is fixed (to be at that 1:1 ratio distance from camera), Z coordinate depends on the 2D platforms Y coordinate (how "high" is the platform).
Once you get the overall picture this gets easy enough, and you could even automate the process to place platforms correctly (but only if you use active objects for these "augmented" 2D platform), otherwise, without a good start setting and/or without paying too much attention, it can be a bit of trial end error XD

I never tried mounting it in an existing game and I guess it would mean: having at least the two "projection" and "map" layers available and named this way, and behind the "front" layer of the game with platforms. Copying over all engine objects (in topleft corner of projection layer), and then the engine group, into the respective layers of the game. Can't remember if the "interface" layer would be strictly necessary, probably not. Then what written above should apply I guess.

--Would using P3D for platforms this way open the door for any new lighting possibilities?

If you think of realtime shadow-casting no, this is unfortunately not supported, would require multi-passing the shader, which afaik is not possible in Fusion. As for straight, simple lighting, how much a 3D face is "bright", P3D provides some "global light" values to tweak the overall min and max brightness of each element in the scene (similar to what you could have with RGB coefficients), and a basic "fixed" shading for boxes (bottom of blocks is darker than top). There's some "more advanced" feature for single meshes, they can react to the lightsource angle and they can react to a "spotlight", you can see it in the "castle example" in P3D, changing light intensity/angle with 1,2,3,4,5 keys, and see what happens when you fire that magic blue projectile or near the glowing torches at the entrance.
Unfortunately the "shapes" shaders didn't have room enough for this feature to be available on them too so the spotlight feature and angle-reaction is limited to meshes, and spheres (which is the shorter shader among shapes).

-- And what would be the options for dealing with objects that are on top of the platforms? Even if the object remains 2d, I guess its x and y position would need to constantly adjust so that the tabletop didn't appear to slide around underneath it? Come to think of it, would Spryke's x and y need to be adjusted too, whenever she's not at the direct centre of the screen (eg. while camera is LERPing, or when at bottom corner of level)?

If the object is not placed too much "in depth" my impression is this is not noticeable and doesn't need to be taken into account, do you see something wrong on this side in that example I posted?
In that example the alien sprite is placed just a few pixel over his invisible hitbox,
this seemed enough for me to make it look over the "depth" side of the platform, but it didn't seem to slide,
since the foremost parts of the platform will be "parallaxing" just a little bit, this is not very noticeable to my eyes... but don't know to better eyes?

That makes sense [MENTION=15682]Volnaiskra[/MENTION] - told you it was a "crazy" idea XD

Just as a last update, I've completed the full template yesterday, and penned down the "blueprints" for the complete starting setup required,
I'll include this example/template in next P3D release but if you'd like to check it out before, just for curiosity, feel free to drop me a pm.

With a simple "auto-platform-creation" code it's literally as easy as dragging and dropping the 2D platforms on screen where you want them on the 2D layer (as you would do normally, basically) width and height are directly taken from the platform graphic (you can even resize the platform directly dragging handles on screen and the 3D block will follow, but this of course will squeeze the platform graphics..), and you can optionally set preferred depth in the platform.
So it's super easy, after a carful starting setup once-forever (that might take an hour),
but only for geometric blocks, as I was suggesting in the beginning.
Those other shapes would require each a custom auto-plaform code and of course more work.

But of course there are other considerations to do for a possible integration in an already existing - and pretty complex - game!

I've included that automatic-platform code too in that example,
I think it's nice so thank you as well for making me dig still a bit more in this

Too kind Vol,
let me thank your skills, insight, accuracy and sincere attitude,
these generously spread qualities all vastly enrich the community
(...and Spryke!)

Sorry you had to spend some money for the engine though, hope you'll have some use of it one day.

Now back to Spryke core works

Maybe you have answered this before, but what program do you use to animate?
I like all the progress posts you make. Spryke keeps getting better and better. Keep it up!

Thanks for the info. I checked the link to Toon Boom and although it looks amazing to work with, it is a bit too pricey for me atm
Keep up the excellent work you do!!

Beautiful... the new platforms really makes it a work of art.

Nice work Vol!.. looks a lot more natural..

I'm impressed with everything you've achieved in Fusion. I've picked up a few of your items in the clickteam shop (or possibly all of them). Can't wait to see Spryke completed.

Very interesting, Vol! That environment does not look tiled at all, I wonder how you tackle this from a level design perspective. Also, good job on your movement engine, handling all those variable surfaces must have been a little nightmare o_O