Requests with ISL!

[juan_irender]

Thanks to you, CTZn.
Some caveats to Ono optimization tips:
1) Using ISL noise instead of the snoise function alters the look of the shader for worst .
2) Using the let..in statement in the r() function has produced an increment of 100 samples/second (PT) in comparison with older function definition.
3) Using a square() function instead of pow(x, 2) has little impact in the performance of the shader.

I´ve uploaded new versión of the shader...

Ciao.

[CTZn]

- String variables in let statements.

It's for convenience only as I'm querying objects from MEL in ISL. Currently it makes the MEL bits more tedious to write, with several MEL evals over a line just to get the coords of the currrently used camera (for LoD notably).

That way I could get the camera name in one statement, then query its pos in another.

- Normalised multifractal output

[CTZn]

crash while invoquing a variable that is declared below. this situation is easily trigged while moving lines around in the ISL editor.

[OnoSendai]

crash while invoquing a variable that is declared below. this situation is easily trigged while moving lines around in the ISL editor.

Can you give some example ISL that crashes?

[CTZn]

it seems to be systematical on my side: declare a variable, use it below, then move the line using the variable above its declaration. I have always been experiencing a crash in that case. Providing such code would be artificial in that sense that there is no way to copy it once the scripting mistake is done.

I will fake and provide such code if you confirm that you could not reproduce the case, thanks.

ps: i've been mostly working on scattering shaders if that helps...

[OnoSendai]

Fixed for next release, thanks for report!

[CTZn]

thanks Ono. 3.8.3 seems picky in certain cases with sliders and render restart but this time I will provide the scene (soonish).

[CTZn]

it's displacement dependent, other parameters do trig a nice render restart.

[CTZn]

scientific notation is usefull in bringing very small or very big values in shaders to the user. however isl does not allow to apply it to params nor variables (may them be bracketed). request is to address these limitations, params being of first importance compared with variables. thanks.

[OnoSendai]

scientific notation is usefull in bringing very small or very big values in shaders to the user. however isl does not allow to apply it to params nor variables (may them be bracketed). request is to address these limitations, params being of first importance compared with variables. thanks.

I'm not sure it's a good idea to allow scientific notation in the params - the parameters are supposed to be easy to use for artists.

Float literals in ISL programs can use scientific notation just fine.

[OnoSendai]

it's displacement dependent, other parameters do trig a nice render restart.

What exactly am I supposed to be looking for in this scene?
What is the expected behaviour, and what actually happens?

[CTZn]

that was related with the ISL editor crash that you got fixed around 3.8.10. Thanks for that !

[CTZn]

scientific notation is usefull in bringing very small or very big values in shaders to the user. however isl does not allow to apply it to params nor variables (may them be bracketed). request is to address these limitations, params being of first importance compared with variables. thanks.

I'm not sure it's a good idea to allow scientific notation in the params - the parameters are supposed to be easy to use for artists.

Float literals in ISL programs can use scientific notation just fine.

OK, here's the catch tbh: Maya will return print values that will read as integers (i.e, 1.0 will read as 1), and appending e0 right after the value will make it a float regardless of it's incoming type.

Also for a moment I thought that using exponents was tidier than multiplying with a very small number but that specific thing is aesthetical.

Your call Ono, no problem.

[CTZn]

in order to clarify: this request is not about UI modification, but allowing the following notations:

Code: Select all

(definedVariable)e-6

and

Code: Select all

userParam()e-6

there was room for this confusion.

[CTZn]

rq: geometry driven depth effects could be achieved with a function that would return the distance to the closest point of the geometry. Meant for volumetrics, as in clouds modeling for instance. There was a request about limiting clouds within a shaped volume.