Maya and ai


#1

well i know ai is a hot topic,

well all artists are faced with a rather brutal realization that without ai assistance,
you arent going to be churning out 500 high quality still variations in a single day for your client to look at,

and your maya renders at slow speed, slow modelling, non parametric stuff, sit next to your competitors ai assisted stuff…not even done in a 3d program at all…but for free access 150 images a day…

i would love to know what autodesks grand plan is to save the artists work life with the current price for illustrations
of $1 to $6 (25picture ai packs on redbubble and etsy…less 50% advert costs).

i think autodesks whole business plan for single individuals, and its non parametric non ai approach
is a total disaster that is happening right now…to the 3d artist

without a 100% parametric approach that is not bifrost,its doomed.

even the current generation of marvel movies images are looking boring to the millions of ai images being
churned out daily…

today a 3d program has to be instantaneous, text promptable,

where is mayas node to switch my building style from giger to tron to final fantasy, when it cannot even loft an even cross section pipe?

the ai tidal wave is here and at least adobe has accepted the fact…has autodesk…

i think even houdini will have to change its maths attribute workflow, its too slow.

quite simply to compete a 3d program has to switch from poly to voxel to nurb to spline patterns on the fly in a totally parametric fashion, auto update uvs at the top of the stack on the fly and go into hires sculpt, and hires paint…sort of houdini…but i can see houdini needs a text prompt …not a maths prompt.

ie you the artist sit down with maya at the start of the day with its one shot poly approach…pay extra for arnold…clocks ticking
at the end of the day you supply your maya only image…versus your competitor in leonardoai or midjourney
who wins?

i know ai image makers are not animation or sculpting programs, well sorry folks, character boards are being done on them, how many views do you need for an auto 3d sculpt to be generated?,…match poses exist…
how many poses are required to text prompt animation?

i am not an ai supporter its taken say 80% of illustration work overnight…ditto retouching and comp.

well all those grand plans autodesk has for maya better appear in its next update,
as i find even blender is way too slow against an ai generator…and blender was way faster than maya.


#2

pity autodesk isnt taking on board the nature of ai,
and the implications it has for its whole business system,

text prompt for maya
fix your nurbs trims, joining curves, build a spline system, build a single unified node system all at the same scale,
build a layered hires sculpt and paint system, build a parametric voxel, poly, nurbs switch.
paint scatter systems, boolean fillets, instance groups with full history,auto remesh, auto uvs at top of stack

negative prompt,
get rid of freeze transforms, double transforms, arbitrary on off transforms in groups, different scales for physical systems, non uniform cross sections along paths, flipped geo along paths, flipping of surfaces when you move nurbs build curves, all its i cant instance that it will fall apart bullshit

ie build a text promttable parametric system


#3

yes i know the attribute box is a text prompt,
but in its current form it quite simply is NOT parametric,
because of the fundamental flaws that existed in v1 maya that were never fixed.


#4

an update on leonardoai,

well lets say it pretends to offer something,
in initial use, you think wow on deeper investigation,
it really only offers the same to all,
you think you have created something unique toyou do a few searches
and see the same demon head, the same dragon, the same insect,
the same space panels, cars…
really only the close up people might be called different,

all comic bacgrounds and styles are really the same.

it is unable to do a basic comp of one creature sitting on another,
and the 2 creatures mysteroiusly end up in a blend you did not ask for.

the res of 1024 is probably less as looks compressed to hell.

the free 150 goes is the minimum you will need to get any sort of low res result.

its like its makes you believe you are being offered an ai image comper when in fact,
70% of your prompt is totally ignored.

so pro illustrators should not really be worried, except for the fact it gives a false impression to clients that see a single close up person in a day, or other robot, that it gacan give you interacting people and objects to fit a brief.

well unless the brief is extremely simple it cannot.

at present it just serves to piss illustrators off, because it makes clients think leonardo ai can do the full job when it cannot,

ai scares the crap out of me, it forces artists to offer styles that are not theirs, in some flash grab way
thats disturbing…

oh for the old days of make a photoshop brush for trees, buildings,
not its the brush of a whole 3d render cobbled from artstation, cgsociety and deviant art

time to become a monk