My initial ideas for continued development to MB3D

[0Encrypted0]

3. Create an interface, where users can create formulas (without much coding involved, especially not ASM).

Any idea of the speed reduction using non ASM formulas?
Wouldn't this also affect the real time navigator?

Parameters that use custom formulas should have a new/different format so they can be shared easily.
Would it be possible to have a plain text parameter (maybe xml = ".m3px") that included the custom formulas?

I would love to be able to search inside a folder of saved parameters for a specific formula or text string.

[thargor6]

Any idea of the speed reduction using non ASM formulas?

This is very hard to predict, the idea is to (pre)compile the formulas. But I did not chose the concrete technology to do this, yet. I will probalby later evaluate some things, and discuss/show the results to make a decision (if there are multiple alternatives to chose from).

[thargor6]

Mutation generation may be interesting

I think this is really an awesome idea and can be developed without too much impact on the other parts of the program. So, maybe, this is also an excellent "warm-up"-addition, I will think about it :-)

[steelsoldat]

@thargor all the best for your ambitious plans!

if you have a real concept for bringing in formula nodes or something, try that but i believe its harder than you might think, because the rendering pipeline is optimised for this, but perhaps you can implement an gpu rendering pipeline, but doing with delphi ... humm ... tricky i think

in my eyes the approach of mandelbulber is the way to go, they already have implemented an gpu based distance estimation module, getting this to be importable by mb3d (because of the more intuitive hybridisation ) would be great but i think it would be more than a year of hard work

at the beggining i would suggest sticking to minor adjustments like more formulas, and a more intuitive formula order editor like one postet here, to be able to swap formulas instead of shifting would already be a great improvement

as well to include totally new formula types like pre-post transforms

but whatever you do do it with fun! i personally underestimated how big the mandelbulb3d crowd is today, we had like 3000 direct downloads on a single day when luca posted the last formula collection with mb3d

I'd like to try and get it to work with OpenCL 

[youhn]

This is very hard to predict, the idea is to (pre)compile the formulas. But I did not chose the concrete technology to do this, yet.

I know Gnofract4D does pre-compile 2D formulas, you can even edit the source text in the same window as the fractals are rendered:



It's open source, so you can look at the method to evaluate.

[DarkBeam]

Looks like a great tip!

[cKleinhuis]

writing a compiler ? uh, that is a quite demanding task and is not done in a single coding session, but if you like

[DarkBeam]

writing a compiler ? uh, that is a quite demanding task and is not done in a single coding session, but if you like

No, but Andreas wanted to

[pulsar69]

That 's a really good news..(further developpement..)

My opinion and wishes.

-keep the actual interface.(my opinion)
- possibility to create some formula is a great feature.
-A second texture slot (one for classic formula, and one for Difs ?), mainly having a different texture for the heigmapdifs will be great.
-A water plane like in mandelbulber.
-possibility to import external gradient. (gradient from apophysis for example..)
- A Possibilty to export 3D object like .obj.  That would be really great !

I probably much reason in artistic way possibility than technical ability.

In any way, thanx to all who works to this project.
philippe

[DarkBeam]

Well, water plane is very similar to gnarly1,2,3 ifs but with more controls.
I don't know if it would be so useful to add another gnarl-type plane?
I am also evaluating to create faster version of gnarlies using (good at 10^-5) approximations instead of slow sines.
To do this I need to replace 'fsin' instructions with a 'call', and also add some constants. Not that long but still time consuming!

[Sockratease]

- A Possibility to export 3D object like .obj.  That would be really great !

Already there!

It's convoluted and requires other software, but MB3D can make a Voxel Stack which can be converted to an obj file.

There are quite a few threads on this topic if you want to see how it's done with examples.

The obj files can be huge and cumbersome  (huger and more cumbersome with more detail!), but worth it if you use 3D Software.

[bib]

Already there!

It's convoluted and requires other software, but MB3D can make a Voxel Stack which can be converted to an obj file.

There are quite a few threads on this topic if you want to see how it's done with examples.

The obj files can be huge and cumbersome  (huger and more cumbersome with more detail!), but worth it if you use 3D Software.

Working the voxel stack through Fiji is possible but not very intuitive. There's a STL export feature in Incendia that simplifies the workflow significantly and offers some settings like size, thickness, algorithm type (cubes, Gaussian...), so why not in M3D?

[3dickulus]

This is very hard to predict, the idea is to (pre)compile the formulas. But I did not chose the concrete technology to do this, yet. I will probalby later evaluate some things, and discuss/show the results to make a decision (if there are multiple alternatives to chose from).

http://lab4.fme.vutbr.cz/heatlab/OpenCLforDelphi.html ?

[DarkBeam]

Working the voxel stack through Fiji is possible but not very intuitive. There's a STL export feature in Incendia that simplifies the workflow significantly and offers some settings like size, thickness, algorithm type (cubes, Gaussian...), so why not in M3D?

Jeremie, in my opinion it's a very hard work as you need to do some complicated 3d matrix postwork with convolutions or similar black magic.
I think it's best to do tiny steps but sure ones

[JMunsonII]

My two cents:

1)  Andreas, GO FOR IT!  Whether you do a re-write, or an extension, have at it!
2)  UI:  Whatever changes get made there, the end result should not be worse than the current.
3)  UI:  Resizable NAVI - Yes!  If you go to a single window preview/render solution, then resizing should automagically take care of itself.  I see why there were two windows - one to render, and one for the Navi.  And with regard to the Navi, I am not overly fond that it doesn't quite show what the render will (even at reduced pixels).  I do understand why it works that way, just not fond of it.  Don't get me wrong, I'm not displeased either.
4)  Formulas:  Having some way to swap formulas in/out (and around) without losing a previous configuration (without doing a lot of file saving/reloading of trial parameter sets) would be great.  For example, I've got three formulas going in the set, I like the Render/Navi result, but I want to try "this" without it being added to the parameter set unless I like the result.  Think "scratchpad."  If the Navi could do that...  If that makes sense...
5)  I would love a GPU render.  Processors can't be swapped out like graphics cards can...
6)  Formula creation:  I'm nowhere near that, but, perhaps some day I could be if I could be taught the math behind it all (and that's a huge challenge).
7)  I'm a visually-oriented person, so, anything that makes the interface work a little more visually is a good thing for me!   This is why I like MB3D over most anything else!
  Lighting:  I don't see why this couldn't be an open-ended thing; in other words, why the 6 light limitation. 
8a) Lighting:  Spotlights!
8b) Lighting/DoF:  Currently the DoF function doesn't play well with the lighting at times (blurs the lighting, especially volumetric) - that really needs fixing.
9)  Formulas:  As with lighting, I don't see why this needs to be limited to 6 either.  But that's really not a big deal, and probably overkill to go beyond that number - but it could happen!

I like the idea of Bokeh and better DoF (maybe add some blur types - Iris, Field, Tilt Shift).

Otherwise, I will take whatever "you all" eventually create (beggars can't be choosers)!

Thanks for taking this on!