Відео

Hey, thanks for watching. 20.5 features a lot of excellent improvements. Adaptive sampling is one of the biggest ones, and that will improve the speed tremendously. Also, the addition of PostFX operations is another nice new feature to Karma. I haven't tested the toon shader yet, so I couldn't tell you how well that compares to Redshift's toon shader at the moment. All in all though, I still suggest Redshift if you're newer because 1. You don't need to confuse yourself with USD right away... and 2. The user experience is still better with Redshift. At some point, it will be worth learning both, but you'll have an easier time starting with Redshift first and then learning Karma.

@@cgforge Super helpul. i using houdini indie as a freelancer, so i really wanted to use karma which included in my license, and supported by side fx directly. but i feel not enough for karma to use. because redshift is sounds/looks better. but i really really hope sidefx will improve karma better than RS personally and imagining your video that 'karma is winner' in the near future.. thanks you kind opinion. sir. have a nice day.

No problem! I would like Karma to win out as well because that would make my life much easier as a teacher. But, it's still not there yet, and frankly, may never get there for beginners unless SideFX focuses on simplifying the user experience. (Which is difficult to do thanks to USD). And, just to re-iterate, I'm talking about beginners - not intermediate or advanced users. I'll be willing to re-visit this video if SideFX dramatically improves its documentation, color correction problems, unifies the interface, and focuses more on user experience. Until then, however, those things are a deal-killer for beginners, and teaching without these things is a nightmare. As an example, take a look at teaching motion blur. I need to explain what motion blur is first (which is a lot for some folks) PLUS all of the weird things that happen when you're working with USD. Same kind of thing goes for explaining attributes. The naming conventions are totally out of wack when, lets say, making a point attribute, converting that info to primvars, accounting for special attributes that are renamed differently during conversion, and then accessing that information through a "USD Geometry Property" node in material X which can be easily confused with similarly named nodes in Material X. As a beginner, you're better off getting a solid grasp of these general concepts first, and then figuring out all the weird USD quirks after you have practice + a solid understanding of basic 3D concepts. So, by using Karma, it's like doubling the workload for beginners to make a good image. For intermediate / advanced users, this may be no big deal, but for a beginner, that could be difference between burning out and sticking with it. So, again, it boils down to... start with Redshift, learn the basics first, and then try tackling Karma. Good luck! - Tyler

@@cgforge thanks tyler, i am really glad to get answer from you really specifically.. i bought RS first depends on your advice. hope you well, and thanks for making videos.

Hey @3DMVR if you're having a hard time remembering everything, then you probably need more practice using Houdini by yourself. It's like driving a car. You need to watch someone else do it, but you also need to do it by yourself to really get the hang of it. Good luck!

I know, my 1080 card has been getting a little senile lately. haha When the new 50 cards come out, I'll jump on that pronto. I have some limited experience with the 30 and 40 cards. The performance is much faster for both Redshift and Karma, but I wouldn't be able to give you specific stats until I upgrade eventually.

Indeed! There are a lot of major improvements coming soon with 20.5. SideFX addressed many issues that were brought up in this video, and it's great to see things change quickly.

@@cgforge could concider making another 1 hour video for that haha

I know right! I'm chasing my tail over here. haha In general though, Redshift still wins out at the moment. Karma is much closer than before. Adaptive sampling is finally here, (SideFX should have released XPU with it because it's such a critical feature when rendering), and PostFX is also a win... but XPU is still not there yet when it comes a few other things. Forcing beginners to learn USD right away is still a pretty treacherous task, there are still many quirks to transparent materials, color management may or may not be better, and the user experience portion is yet to be seen. I'll wait another release cycle before doing an update video. For now, I need to focus on what to do with the, Node "Bible." SideFX decided to make their own node "book" that's nearly identical to the "Bible" that I've been developing for years, and I need to make sure that I'm providing value that goes beyond the SideFX documentation / free videos in order to serve my subscribers.

You can do most of this with the pyrosolver sop. The pyrosolver sop is just a wrapper that goes around a DOP network. So, for the parts that require some custom dop operations, you can go in there and do the same.

I hear ya, but that doesn't really address any of my arguments though. Even if I wasn't selling courses, I would say the same things. In the future, if I truly believe that AI is going to replace the 3D artist, then I'll teach AI to those who aspire to make film / tv / video games / etc. because that's going to help them achieve their goals. But, for now, that's not happening because I don't see it that way.

You'll find mixed opinions on that question, so you'll need to ultimately answer that for yourself. I personally don't believe that AI will replace demand for 3D artists. AI is incredible at generating general ideas + stock footage, but it's terrible at providing customized results to the degree at which clients demand it. Just as an example, if stock footage could replace movies / tv shows entirely, then why hasn't it already done so? CGI has always thrived on creating never-before-seen ideas in a customized, consistent, specific way. None of that lends well to how AI works. But, like all things, time will tell. You'll have to either take the risk by spending your time and energy on it... or avoid it and do something else. Just make sure that it's your decision and not someone else's decision on what to do with your life.

Thanks for sharing your thoughts @hyperman9506 - I don't find your comment to be negative. I think it positively contributes towards the conversation. One thing that ought to have been emphasized more in my video is that there's a huge difference between the concept art / previs work and CGI. Concept art has definitely been disrupted with AI, however, VFX / 3D Animation / CGI is an entirely different situation, and it won't be nearly as affected as concept artists will. In order for AI to replace CGI, it will need to jump leaps and bounds in progress (much like a regular car progressing into a flying car) in order for that to happen, and I just don't think that's in the cards for that to happen anytime soon.

@@cgforge To be honest I don’t see the reason why. Personally I think every think will come down to correcting/polishing the stuf tha Ai did. Why would you pay someone who does it for days if you can get 80 % of the work with a prompt ? Just a matter of the right software coming around.

Well, why not make a movie out of stock footage? That'll often get you 80% the way there. It's because that last 20 percent requires complete control in order to meet a client's demands. Customization is what the clients pay for, and, unlike concept artists, it doesn't need to be mostly what they want - it needs to be exactly what they want. Plus, it needs to move and integrate properly at every frame... with perfect integration in the rest of the scene. You can't just "comp that." And, if you're doing characters, it needs to talk. And... all of that needs to be customized in the exact way that client wants. If a client wants a mirror in a room to be moved somewhere else, as an example, it's not just the mirror. It's the shadows, reflections, shading that reacts to the light in the room, and all the other specific details that will be noticeable once they change a little bit. For clients, that lack of consistency will become very frustrating because there isn't control over anything. Each time you ask for something, it'll be like rolling the dice and hoping that it works well. My point is, AI is good at giving you general things, but it's terrible at giving you anything specific to the degree at which clients demand it. So that's one reason why it has a long way to go in replacing the work of 3D artists.

@@cgforge To be honest I think you are underestimating how fast this is moving. It's one thing to move a mirror with shadows in a program with all the technical skills and a second to do it with prompts and then mabey tweak it a bit wich some sliders ( ex "character maker" ) Sure there will always be someone to art direct the process. But if everything will be prompting based and mabey some very basic sculpting software ( like nomad ) then all those skills ( technicall / software knowledge ) will be worthless. This will lead to the decrease in the perceived value of the role as well as a decrease of people neeeded to do the job. A huge part of the job is knowing the programs and this will be gone if you can say what you want, get it and tweak it. You are no longer working from scratch, you're role is less crucial so less paid. Making a movie form stock footage is not the same, with AI you tell it what you need, just like a client tells a 3D guy now. And again, tweaking sure, but it won't be the same job. It won't be as valuable because you will have to know less to do it. Hope you are right dude but seems like wishful thinking.

I suppose we'll have to agree to disagree about it. Innovation has, indeed, been at breakneck speed. However, just because it has been fast to develop in the past doesn't mean it can continue doing so in the future with unlimited upside to replace all media. I suppose only time will tell. Thanks for adding your thoughts to the conversation.

Nah, I don't think so. It's not really close to replicating what Houdini artists do at the moment. It's a much bigger deal for concept artists, copywriters, and musicians, but Houdini artists are quite safe at the moment in my opinion.

@@cgforge good, i kina have the feeling that u always need pros involved to train the Ai, meaning to get a good music Ai u need musicians looking at its development etc. so as long as no advanced vfx cgi artists snitch and join Ai companies they wont figure it out alone... great videos btw. will inhale them now haha, i know blender very well, curious how long it takes to be good in houdini!

Hey @dancingbetweenus , have you tried changing the viewport lighting options? You should see a variety of small light bulb icons to the right of your viewport.

Oh yeah, that'll work perfectly for getting started. Believe it or not, some of those specs are better than mine right now. For fluid sims, however, you'll want to upgrade your ram up to 128 GB and a bigger SSD / External hard drive. Fluid sims can take up a lot of space on your ram, and, even after you clean up the data and cache to your hard drive, it can take up lots of space.

@@cgforge I definitely can do that! Thank you for your speedy reply... Really means a TON!

@@cgforgeGot my 128gbs of RAM in... Soon I will get a new M.2 4TB SSD... However I will need a power supply soon... Later I'm looking at AMDs Ryzen 9 series CPUs... Looking fast! AMDs R9 16 core cpu smokes the Intel's i9 14900 series...

I think Eevee is worth considering for beginners, but it'll struggle once you start getting into the weeds. The features aren't there compared to the other engines, and migration between Houdini and Blender can get dicey in complex scenes.

@@cgforge With the introduction of Geometry nodes, Blender has closed a significant gap with respect to Houdini. The critical issue that is paramount to developing one's skills, is the ability to quickly iterate, and quite honestly, Blender wins hands down in this respect. The introduction of Eevee Next will address some of its shortcomings, including real displacement, but even now there's Cycles to lean on when non-biased path tracing is required. With the introduction of AI-driven denoisers, I find I can typically get away by setting a relatively high noise threshold (0.1) and allow the denoiser to do the rest of the work, for acceptable results (especially for animation, where it's more forgiving than a still that'll be more closely scrutinized). Taking this into account, I find I can get satisfactory results in under 30 sec for an HD frame even with complex scenes with refraction and subsurface scattering. But the huge point in its favor is productivity. For a student, the best way to accelerate one's progress is to tighten the iteration loop from conception to execution and quite frankly, in this regard, Blender wins hands down. That skill can close the gap with the big boys (Maya, Houdini) has been amply proven by artists like Ian Hubert and Chris Jones, who have shown that it can scale well in capable hands producing professional results that are indistinguishable from its more expensive competitors. But the issue of price, as you have shown, is kind of moot, the true hidden cost to consider is time, not just for rendering, but for quickly getting results that one can respond to and in this regard, Blender shines as it's designed to be accessible and user-friendly for the novice. The Blender road map is quite impressive and the developers have shown that they understand how the tools are being used by the community, prioritizing substance over flashy features - for that there's a rich Addon market. They listen and consistently deliver on features with impressive speed. Even Manuel, from Entagma, who is a staunch Houdini advocate, extolled Blender's virtues in one of his "nerd rant" podcasts and is himself using Blender with his university students, precisely for the reasons I mention. For personal projects, I have recently made the shift to Blender from Houdini and won't look back. It's just the right tool for a lone artist and his laptop. Houdini is unparalleled for working within an established pipeline, but I don't think the lone 3D artist is the ideal fit.

Yeah, ya know, it all really depends on what you're trying to do. Houdini has a way of rewarding your efforts later on in your learning journey and blender does so towards the beginning in general. You will run into a significant number of limitations with Blender, however, and that can turn into big problems real quickly if you need to get exactly what you want at the highest possible quality. I often suggest people learn both because they often have opposite strengths and weaknesses. So, rather than looking at it as 100 percent this or that, it's often more interesting to combine the two depending on the situation. Another thing to consider is that If you render with a 3rd party engine like Redshift, you can use that engine consistently across multiple platforms. One of the problems with native render engines like Eevee and Karma is that you don't get that cross platform compatibility. That can make a significant difference if you're, let's say, freelancing with multiple pipelines to contend with. In that situation, you'll want to learn all the render engines, but it will probably be easier to deal with a 3rd party because of the cross platform compatibility. For example, I've run into a cinema pipeline that uses redshift and a Maya one that uses redshift. In both cases, I'm able to just do all my work in Houdini by exporting out proxies because it didn't really matter where everyone else wanted to be. But anyway, to make a long story short, I wouldn't recommend drinking the Eevee, Karma, or Redshift Kool aid and only doing that. The best answer really just depends on the specific situation, and a great 3d artist is going to learn as much as possible to provide the best custom solution for each situation

Excellent advice!

You're right - next time I'll bring in the air horns, a thousand zoom cuts, and grow a fro mullet.

@@cgforge lmao

Lots of practice! Thanks for watching

See any idiot can use ai 😂

Yep! That's on the agenda for fluids II. We'll see if I can get to it by the end of the year, but it'll be there eventually

@@cgforge will be buying it for sure!! Can't wait

Hi @artadecram_9624 , yes it does. However, I suggest watching the newest version of Houdini For The New Artist here: ua-cam.com/video/IlosUT6_YbE/v-deo.htmlsi=tPL9Sn7hjLB65cU2

Hello @cgsudo , yes, you can bring your FX into Unreal and render it there. That will be good for a game dev / real time portfolio. For VFX, however, you will want to use a full-blown render engine to achieve the highest quality possible. That means using Redshift, Arnold, Vray, Renderman, or Karma for example. Good to know about the Indian wages as well. I had no idea they were offering that little for Jr. FX artists in India. I simply looked up the average Indian wage and assumed that VFX artist would at least make the average. Also, it's worth keeping in mind that what studios offer is not what they might be willing to actually pay artists. In other words, what the studio wants is not always what they are willing to get depending on how well other artists negotiate their wages. It would be interesting to see how much they actually pay vs. what studios want to pay.

I agree with you in general. Much bigger issue has to do with market forces in general rather than AI. Hollywood has definitely been over corporatized and resulted in movies that many people don't really care about as much as they used to. It also wasn't terribly long ago that the Netflix revolution changed the business model entirely for movies, and that's resulted in everyone trying to undercut each other and creating a more chaotic market that's less fun for viewers to sort through. On the plus side, I do think there's a lot of upward potential in episodic TV that uses lots of CG. Game of thrones was obviously a tragedy with how it ended, but it did give a glimpse of what could be possible if you had long format stories that didn't get ruined working with CGI. I suppose only time will tell where the new opportunities will sprout up, but in the meantime, it is definitely resulted in less work for many people in the industry