V4 is scary good, i just run a quadremesher on meshy generated ones, lower the polycount, bake the textures to lowpoly ones, and it is already usable as a prop... 10 mins of work... can't imagine v5. People give too much credit to topology these days. It was pretty important back in the day, not so much today...
Thank you so much for testing Meshy-4! Glad you liked it, and your feedback is invaluable. We are working hard right now on improving textures (hopefully we can get the pupils right, haha), as well as better alignment of 3D models generated against the input image. Another thing we will fix very soon is the tiny floaters in the air. Thanks for the review, stay tuned for the more improvements in the coming weeks!
Yes plz introduce new things in meshy where like we do in real 3d prgrms when we put 4 different angles Front Back Top and Bottom pose for modeling same thing give meshy now with AI plz.....? It will make meshy so perfect. lso plz introduce simole animation tools with correct rigging system too... pkh??
Yeah.. still a way to go to be perfect, but think it's the current leader. Plus I reckon it's good enough for indie games, fun social media videos, or background assets in certain animations.
I agree but it is improving, just like all the AI generators did or are doing. Although I do need to try with a much higher res image, to see how much that improves the results.
@@AIAnimationStudiogood enough for indie games? I mean as place holders for earliest alpha build maybe, if anything more I'd say you have quite a low bar for what you consider quality stuff.
@charredorphan779 will be sufficient depending on the type of art style and game play your going for. Birds eye level rpg type would be far away enough to work.
As a 3D artist, this probably only speeds up the blockout phase of a concept. You still have to build up the model from scratch, but this probably solves some orthographic problems that are apparent when given a concept drawing that doesn't show much of the asset from different points of view. So in conclusion this only speeds up your work by an hour or less.
As a game developer i am more concerned about poly counts for objects and poly shapes for facial. Because its very complex things to edit. Texture u can resize but high poly is notorious for performance issue
@@tomgirl366 . Hey, in their latest update (since posting this video) they’ve added an option for a target poly count. 3k, 10k, 30k and 100k. I’m yet to investigate further but is a small additional feature which may be useful and speed up any manual mesh refinement i.e. on the topology later on.
@@AIAnimationStudio they should add 500 poly also , a simple table can be modelled under 500. I just have checked ,texturing still need to improve as accuracy still bad. Not buying their plan now unless meshy 5 version bring new update.
@TheIndieGamesNL of course there are various technique indeed,but website is still new so giving feedback because it's offering poly count when generating. And it's take time in blender too for unoptimized model with non connected loops or vertices to replace with connected quads. Specially for organic object . But for non organic object like furniture its fine.
Each of these AI tools are great in their own ways. I don’t think it’s ready yet but I do see a pipeline being built around AI tools. Meshy 4 feed into you Retopology AI. Maybe these companies should work together.
Yeah definitely fits that use case. At least as a quick interim model i.e. for a pitch/animatic/placeholder before a proper scan or model could be created.
I appreciate your efforts in keeping everyone informed about the latest advancements in 3D generation technology. However, I believe it's crucial to highlight that these tools, despite being labeled as 'AI,' are still far from being production-ready. The amount of cleanup and remodeling required makes them impractical for professional use at this stage. These videos can be misleading to potential clients, who may develop unrealistic expectations about the value and cost of 3D production. I've encountered situations where clients challenge my time estimates, often bringing up 'AI' as a supposed shortcut, and even consider looking elsewhere due to the misconceptions these videos create. While this might not be a widespread issue yet, it's certainly causing unnecessary friction for some 3D artists who are now facing unwarranted accusations of being overpriced.
Thanks for the comment. I agree... and will try and flag that balanced view more in future videos. The various AI model gen tools have a place, and various use cases (pitches, animatics, placeholder, background objects, references, potentially a base that can then have the topology improved). But agree, for final production the use is much more limited.
Retopo is actually part of the ai model features for most or them now. Rigging is also automatic like mixamo unless placing a few dots is too much labor for you.
I am looking to be a solo game dev. I've tried modeling in the past and now i have options to do it fast and get to the parts of game development that i like.
... buying a graphics card is probably a good idea. Could help with running some excellent 3D software locally, whether it's cascadeur or some local AI generative program. Buy or dont but one of my courses... I don't mind. :-)
Most of these are not very good and the first shoe was totally unusable in my view, it is starting to get good at making stylized Characters and I was impressed by some of the results, but ultimately why would anyone pay the premium Meshy is asking for when there are asset packs made by real artists that look better?
To be honest people shoud stop supporting an promoting this @#&&@ as this will never replace humans creativity and humans idea as you alredy need to give it idea. Its not only 3d skill that makes a 3d artist. So please stop promoting this as people are dumn and think something that takes years of practice to master they can do with ai.
@@synaestesia-bg3ew Exactly what i said. word to word without any room interpritation. IF AI model or a program is not avalible as a free model for anyone to install localy on their own pc, if it has guard rails (cencorship) or is paid or needs access to the internet. if any of the above, even a single statement are true then it is in fact a garbage and should not be given any attention what so ever.
Bad bad AI 3d generated, cannot use in production pipeline at all, I myself can model those with much finest quality in a shorter time. its a waste of time to refine the AI generated 3d model.
This is just the beginning, it’s like the first Atari, of course it’s going to have its limitations right now, but it will be amazing really fast, I feel sorry for your work man, I think is going to be gone as an industry soon, but I think man-made things will always have a special place in our hearts and taste.
@@luisbernal7689 Mark my words, AI will never replace 3D artists who possess real skills. I feel sorry for those who lack the necessary skills and have to rely on AI. It’s not bad, though; it’s like instant noodles-cheap and fast. Chill man, no worries at all
@@SculptEon I will mark your words, now mark mine AI will replace almost everything as an industry, human-made things will become like hand-made stuff, it won’t be necessary, you will buy it because you like it. Right now is a tool that requires skill to excel at a it, it can be helpful to generate easily random stuff but to master it requires skill as any other activity, specially if they require knowledge. Saludos
@luisbernal7689 Even if he's right, there are writers/storytellers salivating for the arrival of when the whole process is easier and quicker. Even if I can never do what he can, I'll be able to create stories that animators can't. The market will be saturated, and storytellers will reign supreme because you can create the whole animation by yourself. Can't wait!!
V4 is scary good, i just run a quadremesher on meshy generated ones, lower the polycount, bake the textures to lowpoly ones, and it is already usable as a prop... 10 mins of work... can't imagine v5. People give too much credit to topology these days. It was pretty important back in the day, not so much today...
Thank you so much for testing Meshy-4! Glad you liked it, and your feedback is invaluable. We are working hard right now on improving textures (hopefully we can get the pupils right, haha), as well as better alignment of 3D models generated against the input image. Another thing we will fix very soon is the tiny floaters in the air. Thanks for the review, stay tuned for the more improvements in the coming weeks!
is it possible to use multiple images from different angles to create a 3d object?
Yes plz introduce new things in meshy where like we do in real 3d prgrms when we put 4 different angles Front Back Top and Bottom pose for modeling same thing give meshy now with AI plz.....? It will make meshy so perfect.
lso plz introduce simole animation tools with correct rigging system too... pkh??
One thing I wanna test out is generating individual parts of one object to see if it reaps better results.
love it, they were good before, now its wonderful
But the amount of cleanup work needed to make these things usable is still pretty high, both at the topology level and especially the texture level.
Yeah.. still a way to go to be perfect, but think it's the current leader. Plus I reckon it's good enough for indie games, fun social media videos, or background assets in certain animations.
I agree but it is improving, just like all the AI generators did or are doing. Although I do need to try with a much higher res image, to see how much that improves the results.
@@AIAnimationStudiogood enough for indie games? I mean as place holders for earliest alpha build maybe, if anything more I'd say you have quite a low bar for what you consider quality stuff.
None of these are ready for real production. Most would be for solo indie game devs or film makers or hobbyists which is still alot of folks.
@charredorphan779 will be sufficient depending on the type of art style and game play your going for. Birds eye level rpg type would be far away enough to work.
As a 3D artist, this probably only speeds up the blockout phase of a concept. You still have to build up the model from scratch, but this probably solves some orthographic problems that are apparent when given a concept drawing that doesn't show much of the asset from different points of view. So in conclusion this only speeds up your work by an hour or less.
I am impressed how well text to model did work
As a game developer i am more concerned about poly counts for objects and poly shapes for facial. Because its very complex things to edit. Texture u can resize but high poly is notorious for performance issue
@@tomgirl366 .
Hey, in their latest update (since posting this video) they’ve added an option for a target poly count. 3k, 10k, 30k and 100k. I’m yet to investigate further but is a small additional feature which may be useful and speed up any manual mesh refinement i.e. on the topology later on.
@@AIAnimationStudio they should add 500 poly also , a simple table can be modelled under 500. I just have checked ,texturing still need to improve as accuracy still bad. Not buying their plan now unless meshy 5 version bring new update.
@@tomgirl366 . Yeah, I definitely agree; being able to choose even lower poly counts would be really useful for a lot of simpler models. 👍
you can just lower polygon definition in Blender
@TheIndieGamesNL of course there are various technique indeed,but website is still new so giving feedback because it's offering poly count when generating. And it's take time in blender too for unoptimized model with non connected loops or vertices to replace with connected quads. Specially for organic object . But for non organic object like furniture its fine.
Each of these AI tools are great in their own ways. I don’t think it’s ready yet but I do see a pipeline being built around AI tools. Meshy 4 feed into you Retopology AI. Maybe these companies should work together.
A fair start, But, Woof. Still a lot of Clean Up to do on most of my results.
Decent for 3D references, similar to a 3D scan
Yeah definitely fits that use case. At least as a quick interim model i.e. for a pitch/animatic/placeholder before a proper scan or model could be created.
can we use these 3d modeles in maya?
sure
I appreciate your efforts in keeping everyone informed about the latest advancements in 3D generation technology. However, I believe it's crucial to highlight that these tools, despite being labeled as 'AI,' are still far from being production-ready. The amount of cleanup and remodeling required makes them impractical for professional use at this stage. These videos can be misleading to potential clients, who may develop unrealistic expectations about the value and cost of 3D production. I've encountered situations where clients challenge my time estimates, often bringing up 'AI' as a supposed shortcut, and even consider looking elsewhere due to the misconceptions these videos create. While this might not be a widespread issue yet, it's certainly causing unnecessary friction for some 3D artists who are now facing unwarranted accusations of being overpriced.
Thanks for the comment. I agree... and will try and flag that balanced view more in future videos.
The various AI model gen tools have a place, and various use cases (pitches, animatics, placeholder, background objects, references, potentially a base that can then have the topology improved). But agree, for final production the use is much more limited.
kinda expensive for indie, 1 model takes around 25-(USD 1) - 75(USD 4) coin.
which one better vs rodin ai?
Bro knew if he put that in blender, blender braking his pc in return😂
I have created a character but it has many imperfections (holes) what can be done? please give some promo codes :) help
Gen AI devalues all the creative parts of work doesn't it, still leaves the retopology and rigging for the humans.
Retopo is actually part of the ai model features for most or them now. Rigging is also automatic like mixamo unless placing a few dots is too much labor for you.
It allows anyone to create now. The less limitations there are the more opportunity there is for people to create. It also saves a lot of time.
I am looking to be a solo game dev. I've tried modeling in the past and now i have options to do it fast and get to the parts of game development that i like.
why buying ur course when i can buy a better grafics card ai dose it for me anyway
... buying a graphics card is probably a good idea.
Could help with running some excellent 3D software locally, whether it's cascadeur or some local AI generative program.
Buy or dont but one of my courses... I don't mind. :-)
Can you use offline?
No, it's all via their site.
Creativity is now like fishing or treasure hunting..
Bro is there any way we can make our 3d model character move or lipsing from our video for free. I need an avatar for my UA-cam channel
The topology is worst thing I've ever seen
Is it free?
I hope in the near future, Meshy will be able to produce photorealistic characters with realistic hair.
Herkese zengın olma dılerım
Nice models, but it's neigh unto useless for learning Meshy.
Show us the mesh, those models are useless when the topology is fucked up.
Most of these are not very good and the first shoe was totally unusable in my view, it is starting to get good at making stylized Characters and I was impressed by some of the results, but ultimately why would anyone pay the premium Meshy is asking for when there are asset packs made by real artists that look better?
To be honest people shoud stop supporting an promoting this @#&&@ as this will never replace humans creativity and humans idea as you alredy need to give it idea. Its not only 3d skill that makes a 3d artist. So please stop promoting this as people are dumn and think something that takes years of practice to master they can do with ai.
As long as you are unable to run it locally, with 0 guard rails, for free and no connection to the internet then the model doesn't exist.
Simple as.
What do you mean?
@@synaestesia-bg3ew Exactly what i said. word to word without any room interpritation. IF AI model or a program is not avalible as a free model for anyone to install localy on their own pc, if it has guard rails (cencorship) or is paid or needs access to the internet. if any of the above, even a single statement are true then it is in fact a garbage and should not be given any attention what so ever.
Bad bad AI 3d generated, cannot use in production pipeline at all, I myself can model those with much finest quality in a shorter time. its a waste of time to refine the AI generated 3d model.
This is just the beginning, it’s like the first Atari, of course it’s going to have its limitations right now, but it will be amazing really fast, I feel sorry for your work man, I think is going to be gone as an industry soon, but I think man-made things will always have a special place in our hearts and taste.
@@luisbernal7689 Mark my words, AI will never replace 3D artists who possess real skills. I feel sorry for those who lack the necessary skills and have to rely on AI. It’s not bad, though; it’s like instant noodles-cheap and fast. Chill man, no worries at all
@@SculptEon I will mark your words, now mark mine AI will replace almost everything as an industry, human-made things will become like hand-made stuff, it won’t be necessary, you will buy it because you like it. Right now is a tool that requires skill to excel at a it, it can be helpful to generate easily random stuff but to master it requires skill as any other activity, specially if they require knowledge. Saludos
@luisbernal7689 Even if he's right, there are writers/storytellers salivating for the arrival of when the whole process is easier and quicker. Even if I can never do what he can, I'll be able to create stories that animators can't. The market will be saturated, and storytellers will reign supreme because you can create the whole animation by yourself. Can't wait!!
@@SculptEon“mark my words!“ people in 1815 when they said steam engines would never replace horses
can we use these 3d modeles in maya?