@@Sneguro4ka Thank you so much. I'm glad you find it useful.

@@kirigayakazuto1997 I would say can the blender export settings accordingly in blend file. By default it JPEG/PNG and if you want exr then select appropriate options. Rest everything I said in this video still applies.

@@AlEgorova set withCredential property to true in frontend API library like axios or whatever you are using. That way you can parse the cookie value. More info - flaviocopes.com/axios-credentials/

@@ZacharyGarner-e9r try saving the blend file with eevee as a renderer and see it's working or not.

@@RahulAhire Do I need to specify EEVEE instead of Cycles in runpod?

@@ZacharyGarner-e9r I mean in the blend file. In the Blender docs, I didn't found anything specific to EEVEE i.e meaning I've to test and then update. Right now, I've updated the scripts to use OPTIX as a renderer so you can get great speed and GPU is 100% utilised all the time. So, if you can then please try using CYCLES as option in blend file.

@@RahulAhire I understand. Cycles is alot better for rendering. My computer is not good however and I usually do everything in EEVEE. But I am so happy you made this it helps me out alot.

@@techEnthusiasm369 it has some minimum quota as free tier for frontend. You can check their pricing page.

@@himanshuvaishya9164 Thanks for watching this video Himanshu, I hope this presentation helps you any shape or form. Please do checkout the demo on your laptop. Link should be in description.

@@kewlorsolutions I haven't worked with Unreal so I can't tell much in detail. You can rent a Windows GPU VM with other providers to get a remote desktop like GUI environment. But that's little technical.

@@mondei1 As devs like you and I, it's easy to get carried away by thinking about adding more features to make it essentially look good. I'm sure you can also make an UI with gradio but handling errors for all its case is challenging. Someone at Runpod actually made an UI after watching this tutorial but it wasn't working as expected. I wanted keep this tutorial as beginner friendly as possible so that even 3D artists can understand what's happening here. But yeah, thanks for watching the video. If there's anything else I can help, please let me know.

​@@RahulAhire I actually tried the script on an instance but it got automatically killed by RunPod at the "Building BVH" step. RunPod showed me a 100 % RAM usage so I guess this was the problem. It was a 4090 pod with the 4.1 splash demo. Maybe I was just unlucky. But someone working for RunPod did something for Blender? Crazy. Anyway, I agree with the instant drift to make something out of an idea 🥲. You get peak motivation at the start but this quickly fades away after some work. But my naive brain can't think of anything complicated in this idea. I will come back here when I have finished something. Could be in a few months or never :p

@@mondei1 try other Runpod instance. Sometimes I've seen that render can be freeze but it only happens few times. If your blend file is really big (in GB) then it can take time. Make sure you can adequate RAM. Sure, I'm always here 👍🏻.

@@amandhimank7707 thanks 🙏🏻, I hope it helps you in your project

@@Alphain you mean the file format exr instead of png? Try changing the settings in blend file and also the script line where it mentions .png to .exr

@@RahulAhire where can i contact you?

@@Alphain received your DM on IG

@@Keth-it9xb ideally you shouldn't do that but since I don't know your usecase I'll say that create an API when sending the post request to it with cookie using withCredential return the response as a json value and then read it.

@@santipriyabodeboina1058 Hi there, may I know on what area of this video you're confused about? I'd love to clarify...

@@williamfernandez2185 this is the code, I don't know what's else you need. Can you please clarify?

@@RahulAhire claro,, es para utilizar este codigo en kaggle,, es un servivio gratuito de cpu y gpu,, para renderizar con blender pero en la plataforma de kaggle. gracias

@@williamfernandez2185 if kaggle support jupyter notebook then download the script given in the description and follow the tutorial. It should work fine as intended if kaggle has already configured CUDA.

@@Vivavisuals3D the simple answer is no because both of them don't have command line interface version like blender. If you want you can spin up a window based GPU GUI on Tensordock if that's what you needed.

Also, how do I do it, when I have a video as a shader.

@@tommyleejohn thanks for coming by and watching the video. The default settings are such that you don't need to select any of them. The script utilises both CPU and GPU to its full potential. Edit - I haven't worked in shader so could you explain what does that exactly mean?

@@RahulAhirehey okay Perfect! In shader would be I have a Screen where I put in a video for example. The shader is the material that has the video inside with a Image Texture node. So the video file is outside the blender in the finder structure

@@tommyleejohn you mean video texture? For that keep the video in the same folder where the blend file is. And keep the same in the jupyter notebook as shown here. Edit - don't forget to link the video file again file uploading the blend file.

@@CipherEspanol can you expand on what type of plugin you're using that depends on this rendering process?

@@RahulAhire let’s say I want to use GeoScatter, but I don’t want to bake the instances it creates. How can I install that plugin in the cloud? Sorry for the questions but I am a beginner.

@@CipherEspanol thanks for detailed context. I just wanted make sure I understand in the same way as you do. So I find out that you can unzip the addon in the addon folder and just configure your blend file properly locally. Have a look at this answer - blender.stackexchange.com/a/73760 I hope you can be able to locate the folder in this jupyter notebook as I've shown in the video. Let me know if it helps. PS: I did saw your printer hacking video in my recommendation few days ago but I watched it right now and it's really good I wonder how many months it'd have taken to create it. Are you doing it all by yourself or you're having a team?

@@RahulAhire Oh really!? Thanks a lot brother, very happy you liked it. I (Bored) did all the video myself. We are 2 guys running the channel at the moment. The video took me more than 200h, but I hope I can go faster in the upcoming videos, lol. Btw, will try what you said and see if it helps. I also followed you on X ;)

@@CipherEspanol Oh, I do understand it's tedious to make video. This video of mine - The man who invented Trillion Dollar Algorithm : kzbin.info/www/bejne/h3WxY4WbYtqjgpI took about 4-5 months from research to press that upload button. Anyway, I've sub'd your channel, looking for to learn your cinematography style and see how in what way I can implement them on my channel. Edit - if you want Windows RDP (GUI) you can try platform like Tensordock.

In production, have a common base domain for eg. the frontend is google.com and the backend might be api.google.com and rest all the principles follows the same as I've shown in the video. How to send them? - if you are using axios then use the with credentials settings to attach the http cookie on each request you send to the server. flaviocopes.com/axios-credentials/ Edit - you're right http cookie are not accessible by javascript. You can still view them in cookie tab in dev tools only for yourself just like how you own the keys of your home.

Thanks for the reply, I think my problem is that I'm using amplify authenticator which gives the tokens to the client and I can't create them server side as this library is client side only.

@@daviddoyle7580 well see if you can send those token from backend sort of proxy via set cookie method. Other than that, I don't know how cognito specifically works so can't tell anything about it.

@@RahulAhire Thanks for the reply, I think the tokens should just always be received initally on the serverside for safety, I dont even understand why AWS made this Amplify client side library in such a way that its xss vulnerable.

You'll have to send a clear cookie request from server. The principal applies the same. Also make sure you do take care about timezones since server are defaulted to UTC.

@@RahulAhirei have tried everything unfortunately the exact same code works in local environment but not in hosted server. And what to do with time zones??

@@edwardgaming5681 without much details I cannot answer much in detail. Can you exactly tell me what happening in detail?

Could you please provide more details? Have configured you cors?

@@RahulAhire yes i have configured cors app.use(cors({ origin: `${process.env.CORS_URL}`, methods: ['GET','POST','PUT','DELETE'], credentials: true })); is it right?

i am using fetch

@@bpravin7807 anything more specific than that?

With this setup - the answer is no because it's a Linux based system not a windows or Mac. Now you can use windows container but it's technical and not suitable unless you have a dev know how. My suggestion would be look for a cheap windows GPU VM (like Tensordock or many are such available) and just RDP it and install the clo 3D.

@@RahulAhire tks. could you please recommend a tutorial in youtube to do it ?

@@scmusicacademy1341 check this out docs.tensordock.com/virtual-machines/how-to-rdp-into-your-instance

Thanks for watching the video, I really appreciate it. If you follow the standard blender settings, you should be able to render out images. I haven't deal with multi camera setup so I don't know if blender creates multiple folder for different camera. The script works on per scene basis so I'll have to try to properly confirm. Regarding K-cycle, the answer is no, at least on the command line version as demonstrated in the video.

Thank you so much for your reply! This is super useful information!!! Thank you very much for your help! I will definitely check out your suggestion! Thank you 😊

@@andreabell189 if you strictly want GUI based blender setup, you can checkout tensordock where you can rent Ubuntu server and install xRDP (see this guide - linuxize.com/post/how-to-install-xrdp-on-ubuntu-20-04/) on it. It's little technical but if you can manage to get it running then it's basically a virtual local desktop. Only drawback is if you don't live in US or EU then the ping are so high its unusable.

@RahulAhire oh great, I will investigate... I am uk based so might not work..... how would this compare to something like a render farm like fox render... really I am looking for the cheapest and fastest way to render out an animation. If I have to use python or whatever so be it.. thanks again for your help! Your great 👍

@@andreabell189 compared to fox render both Runpod and Tensordock are great. Runpod is quick and easy to use compared to Tensordock which has little additional step. You can use Tensordock from anywhere. By EU, I mean their server are in US, Ukraine and Poland. Runpod is container and Tensordock is Server which gives you more control (given you know what you're doing). However from last few days I've noticing that there's a shortage of GPU and performance is little inconsistent on Runpod. The support team has indeed assured me that it's a temporary issue as they are removing old GPUs and will be resolved soon. Tensordock is slightly faster (5-15 secs/frame). So I think this should help you to make a decision. Edit alerts (8 December 2024) - I just noticed the performance difference isn't entirely the fault of Runpod. I happened to discover how to use optix today and boom, GPU was running at its full speed giving max performance compared to not using optix.

Actually you can - have a look at this blog it shows you how to deploy a RDP Ubuntu server in which you can install blender. There's also a link for the template in the blog. blog.runpod.io/gpu-accelerated-virtual-desktop-on-runpod/

@@RahulAhire TBH I didn't used anything except windows. Right now I am using paper space windows virtual machine at price of $0.52 /Hours. But the GPU isn't fast. It is Quadro P4000. Is there any other virtual service you can recommend? That offers windows

@@BoxingCommunity07 most online services are based on Linux so it'd be hard to get it configured. With that said, blender officially supports Linux so I don't think you'll have any problem in using that.

@@RahulAhire Hmm, I haven't used Linux that's why I am having trouble if I can get view just like desktop where blender shortcut is added. Is it possible 🙁

@@BoxingCommunity07 you can download blender for Linux. See the blender download page for more options. Edit - it's the same gui interface that we are used to in windows. Look for blender in Linux tutorial.

Thanks for your candid feedback, I appreciate it. Can I know what you were looking for? This video is about the conceptual overview rather than actual coding deep dive.

Can I understand what you were looking for in this video? This is an conceptual overview of the JWT access and refresh token.

​@@RahulAhireI still do not understand what is the reasoning behind refresh tokens. You keep it http only, good and great then why not just send the access token as an http only cookie itself? The assumption here is that refresh tokens are http only cookies and will not get stolen. The same can be said for access tokens sent as http only cookies.

@@youdontknowme2508 ok let say you only opt for access token. What if the token expires due to short session, the user will have to do the login again resulting in poor UX. Ideally for security you should have access token expiry for only few hours and let the refresh token issue new access token. By introducing refresh token and a route to handle it, you introduce more granularity and can rate limit on how much access token you want to generate thus enhancing the security.

@@RahulAhire The security reason applies to Refresh Tokens as well. You're keeping refresh tokens as a cookie for a very long time. What if it gets stolen? If it gets stolen that's basically equivalent to your access token also getting stolen by that logic. You're keeping Refresh Tokens as an http only cookie. If that guarantees that it will not be stolen then Access tokens set as cookies also will not be stolen. For the limiting part, this could be still done with access tokens as well since you're anyways maintaining a state on the backend server.

@@youdontknowme2508 security isn't a simple mechanism, you can associate refresh token with device info and screen resolution as a matter of device fingerprinting. Ideally you should keep both tokens with http cookie which prevents XSS. You can use redis or inmemory store to invalidate refresh token. Think about how you might remove Google account from your Android device which Google has highest preference for security access eg. Passkey or web authn. If you want to introduce state in backend then I don't think that's a proper way of using jwt. You should not use JWT like session stored in database. It add additional latency on every request as db needs to do a lookup.

Hey quick question did you have an issue with the volumetrics? My fog isn’t appearing in my renders and I’ve been using the same settings for a long time

Hi Ashley, sorry for late response. I missed the notification. I don't know what your settings ares but in general there seems to have an issue regarding this - blender.stackexchange.com/questions/238685/having-issues-with-volumetric-fog maybe you can try this out. I haven't worked with volumetric fog so I can't tell much about it.

Sorry for your experience. May I know what you were looking for? I'll try to answer it to the best of my knowledge.

The answer is obviously no. It's not safe but this video was about the demo and not web security. You can find various articles and tutorials on securing frontend API keys.

@@RahulAhire hi, thanks for responding. Please can you link me to a tutorial or article that teaches how to secure the apikey. I have tried many methods but I couldn't find a way. I used the method where I will get the script in a useeffect inside a react component but that is throwing an error. Please I'd appreciate a response on this, thank you!

@@olaoluwaanigboro-napoleon6484 I think you can follow this tutorial kzbin.info/www/bejne/kHjcnoFulqumrJY or else you can search for the term "securing frontend API keys" Most often, you'll need a backend server for it.

@@RahulAhire alright, thanks I appreciate. I'll check that tutorial you sent

Well I'm actually focusing on my startup Experientia. You can see it's video right here - KZbin.com/@experientiaYT and read about the mission statement - blog.rahulahire.com/what-is-experientia

Good 👍🏻, I hope it helps you to build secure apps.

eg 6:46 in, and havent started an instance yet :P

Thanks for watching this video. I really appreciate it. I could explain the video in short but it won't make comprehensive sense for the viewers. But I've indeed added the timestamp so that might be helpful. If not then let me know how should I correct it...

Can you give me a brief detail on what you were trying to do and on what API are you getting this error?

@@RahulAhire Sure. Thank you for the reply. I am using my phone so dont have code snippets. I am using python. The setup I have is I have api gateway which calls lambda and lambda generates presigned url for upload and download. I was using generate_presigned_url boto3 function to generate the url. When I used just the bucket, file, uploadId and part number as parameters for the function (no extra headers) I was able to generate the presigned url but when I tried to use that url to upload the part, it returned the "no such upload" error. I was able to get past it by passing in the host header and setting it to my bucket. But now I am getting a signatute does not match error

@@RahulAhire in the video, It did not look like you had to set any extra headers up other than the ETag one. I am wondering if it is because generate_presigned_url does not take a content_type header

I guess it is also important to mention that when I tried same boto3 calls directly from python to s3, they worked. So it is most likely something going from the api gateway to lambda. I do have cognito authorizer set up and it passes the authorization header

@@shababashfaque447 sorry for delayed response, thanks for mentioning your problem in detail. Just wanted to know are you using presigned url for only 1 file or it's a multi part upload? Since you tried to use the lambda, have you checked how the lambda is handling the headers. I think its likely a library issue and I think you can check your network tab to see how the request is being sent i.e complete multipart upload and resolve it accordingly in lambda.

It's coming on @experientiaYT

Well it depends on how you deploy your micro services but fundamental are still the same. You can have unique subdomain for each bff and set the httpcookie to LAX.

Platformjs provide all the details out of the box so that's nice.

@@RahulAhire platformjs? I don't understand... Can you explain better, because I'm seeking to know the device model and screen size for a given user agent without necessarily using DeviceAtlas or WURFL APIs as I don't have the budget for their of the platforms right now.

@@arakunrindamola I meant that it directly give you info on device model from the browser itself. You can jsonify other details into single object. Not sure about what you are trying to extract in user agent string.

@@RahulAhire okay... Like I said, I'm majorly to extract the device model and screen size of a given User-Agent... I'd like to use the screen size details in my traffic bot and I want to be as exact as I can... I don't want to just use any screen size I like... It might leave a loophole in easy tracking of my bot.

@@arakunrindamola so if I'm correct you can get device info from platform.js as I mentioned in the video and that's much better than parsing the user-agent string. You can even try the demo from the link in the description. Regarding screen size you can either way use window.innerwidth and window.innerheight and screen property for screen orientation, etc. So am I missing something?

Thanks for watching it, really appreciate it.

Maybe you should look for colocation alternative like minds of Phoenixnap and others which provides baremetal servers on demand with terraform support. Future is going to be more bright. Maybe you'd like to have a look into DHH video on moving out of cloud. *Sorry for late response. This comment didn't showed up in my notification feed.

Thanks Yash, I appreciate your comment.