A beginner’s guide to Xesktop

What is Xesktop?

Xesktop is a server rental service that gives users access to remote GPU servers that behave like cloud-based Windows computers. Instead of buying or upgrading a powerful workstation, you can rent GPU power when you need it and control the server through a Remote Desktop connection.

For beginners, the easiest way to understand Xesktop is this: your own computer is used to connect, while the heavy work happens on a stronger machine in the cloud. This can help when your local machine struggles with rendering, AI generation, simulations, high-resolution scenes, or GPU-heavy software.

A service connected to GarageFarm.NET

Xesktop exists alongside GarageFarm.NET’s cloud rendering service. If you already have a GarageFarm.NET account, you may be able to use the same login details for Xesktop. The difference is in the workflow. GarageFarm.NET is focused on render farm submission, while Xesktop gives you direct access to a remote GPU workstation.

Who is Xesktop for?

 

Xesktop is built for people who need more GPU power without buying a new machine. This can include freelancers, studios, AI users, 3D artists, designers, developers, and production teams. It is especially useful when a task needs a strong GPU for a limited amount of time. Instead of investing in expensive local hardware, you can rent a server for the session, project, test, or production period.

3D artists and render users

For 3D artists, Xesktop can be used to run GPU render engines and creative applications. This may include workflows in Blender, Cinema 4D, Maya, 3ds Max, Daz 3D, OctaneRender, Redshift, Cycles, and other GPU-based tools. It can help when a scene is too heavy for a local workstation, when deadlines are tight, or when you want to keep your own computer free while rendering happens remotely.

AI and ComfyUI users

Xesktop can also support AI workflows, including ComfyUI on supported machines. This is useful for image generation, video generation, model testing, AI development, and other workflows that need more VRAM. For users who do not have a high-end GPU locally, Xesktop can make heavier AI workflows more accessible.

Studios and production teams

Studios can use Xesktop when production demand temporarily grows. A team may only need extra GPU power during a deadline, testing period, or heavy render stage, so renting a remote workstation can be more practical than buying new hardware. Because each system runs as its own remote desktop, teams can also keep workflows separated while still using shared storage for project data.

How the Xesktop dashboard works

After logging in, the Xesktop dashboard is where you manage your servers, images, credits, and connection details. This is the main control center for starting and stopping your remote workstation. The dashboard also shows useful details such as server availability, credit balance, image activity, lease duration, and cost.

Comparing available servers and other information

From the dashboard, you can browse the available server options by going to the info panel, and clicking “compare” beside the Server specs. You can also use the comparison view to see different servers side by side. This is helpful because different jobs need different types of GPU power. A render job, AI workflow, or GPU-dependent tool may benefit from a server with more VRAM, more GPUs, or a newer GPU model.

You can also find various resources in this info panel. To close the whole thing, simply click the blue text “hide info panel”. To reopen it, click the blue text “show info panel” located below the navigation bar.

Checking credits and availability

The top area (or navigation bar) of the dashboard shows account details such as credits and server availability. Before starting a long session, check that you have enough credits for the work you plan to do.This is especially important before running long renders, AI batches, or production tasks that may continue for several hours.

Checking image activity and creating new images

The area below is where you can find your image activity. On the right side, you may filter the images via All, Active, Inactive, and Preparing. On the left, you can click the “new image” button to create a new image.

How to get started with Xesktop

The first setup takes a little time because Xesktop needs to create your system image and boot the server. Once that is done, connecting becomes much easier.

Step 1: Create a new image

To begin, click the option to create a new image. You will be asked to add a description or name for the image. This name is mainly for your own organization, so choose something related to the project, tool, or workflow.

Clicking YES to “Boot system right after the image is ready?” pertains to automatically starting the machine as soon as the new image has finished being created. 

Step 2: Wait for the image to be created

After you create the image, Xesktop prepares a fresh Windows environment. This can take several minutes, so it is normal if the image is not ready immediately. Once the image is ready, you can go to its card and see its current state, activity, lease duration, and cost from the dashboard.

Step 3: Set the resolution

Before booting the system, you may be able to choose the display resolution for the remote desktop by clicking the three dots beside the trash icon, and go to Set RDP resolution. This helps match the server view to your monitor or preferred working setup. A higher resolution can make the workspace more comfortable, especially for 3D, compositing, or node-based tools, but it may also depend on your local display and connection quality.

Step 4: Boot the system

When the image is ready, click the button with the play icon. Xesktop may show a list of available GPU servers that you can choose from. Pick the server that fits your task. A heavier render, AI model, or GPU-intensive application may need a stronger GPU, while lighter setup work may not require the highest-end option.

Step 5: Download the RDP file

After the system is running, download the RDP file from the dashboard. This file contains the connection details needed to access your remote desktop. You can open it with a Remote Desktop application. On Windows, this may open directly. On macOS, you can use a Remote Desktop app such as the Windows App or Microsoft Remote Desktop.

Step 6: Log in to the remote desktop

When connecting through RDP, use your Xesktop account details. The username may be your account number, which can be found in the dashboard, and the password is the same one used for your Xesktop account. After logging in, you will enter the remote Windows desktop. From there, you can begin setting up the system and working on the server.

Installing software on Xesktop

Once connected, you can install the applications you need for your workflow. This may include 3D software, render engines, AI tools, plugins, utilities, browsers, file transfer tools, and monitoring software. The important thing is to install applications in the right place and keep project files separate from system files.

Use the C drive for applications

The C drive is the system drive. It is the right place to install applications, plugins, and program files because it is designed for the operating system and installed software. You should avoid using the C drive or desktop as your main project storage location. Treat it as the place for software, not as the place for working files or final outputs.

Use the user drive for working files

 

Your working files should go on the user drive or shared storage drive, often shown as the U drive. This is where you should save project files, source assets, render outputs, AI results, caches, and other important data. This drive is also useful because it can be accessed through FTP. That means you can move files in and out more safely without relying only on the live remote desktop session.

Keep your setup reusable

Applications installed on the system image are preserved, so you do not need to reinstall everything every time you boot the same image. This is one of the biggest advantages of using a saved image. After your software is installed and configured, you can reuse the image for future sessions and continue from a familiar workspace.

Moving files in and out of Xesktop

File transfer is an important part of the workflow. You need a reliable way to bring project files into the remote workstation and download final outputs when the work is done. There are several ways to do this, depending on your workflow and file size:

FTP access

FTP is one of the main ways to access files stored on the user drive. This is useful because your files can be reached even when the server is not actively being used. For large projects, FTP can be more organized than dragging files through a remote desktop session. It also helps keep files in the correct storage location.

Cloud storage

You can also use cloud storage services such as Google Drive or Dropbox to transfer files, depending on your setup. This can be convenient if your project files are already stored in the cloud. However, cloud services may need extra configuration, especially when syncing to a remote machine or working with large files. For production work, always confirm that files have fully uploaded or downloaded before starting.

Remote Desktop copy methods

For smaller files, you may be able to copy and paste or move files through the Remote Desktop session. This can be convenient for quick transfers, but it may not be the best choice for large projects or heavy render outputs. For serious work, FTP or organized cloud transfer is usually easier to manage.

Billing and credits

Xesktop uses a credit-based, pay-as-you-go model. The exact cost depends on the server type and current pricing, so you should always check the dashboard before starting a long session. The main thing beginners need to remember is that credits are used while the server is running.

Billing starts when the server is running

Once a server is booted and active, billing begins. Usage may be calculated in small time increments, so even shorter sessions can be billed fairly. This gives you flexibility, but it also means you should be careful not to leave the server running when you are no longer using it.

Billing stops when the server is shut down

Closing the Remote Desktop window does not necessarily stop the server. It may only disconnect you from the session. To stop charges, shut down the server properly from Windows or through the Xesktop dashboard. Make this a habit at the end of every session.

Add credits before long jobs

If you are planning a long render, AI batch, or heavy processing session, check your credits before starting. Running out of credits during a job can interrupt your workflow. It is better to estimate the cost first, add credits if needed, and then start the job when you are ready.

Software licenses

Xesktop provides the remote server, but it does not automatically provide licenses for every paid application you may want to use. If your software requires a license, you are usually responsible for providing it. Before starting production work, check whether your software license can be used on a remote workstation.

Bring your own licenses

For paid tools, you may need to log in with your own account or activate your own license key. This can include 3D software, render engines, plugins, AI tools, or other production applications. When your work is done, remember to deactivate or release licenses if your software requires it.

Check license restrictions early

Some license systems are tied to specific hardware or network setups. These may not work smoothly on a cloud server. If you are unsure, test the license before starting an important project. This can save time, credits, and frustration.

Final thoughts

Xesktop gives beginners a practical way to access powerful GPU hardware without buying a new workstation. It works like a remote Windows desktop, but with cloud GPU power behind it.

As you become more familiar with the workflow, Xesktop starts to feel less like a separate cloud service and more like an extra workstation you can open when your local machine is not enough. The key is to build a clean routine, keep your files organized, test your setup before committing to heavier jobs, and shut things down properly when the work is done. With those basics in place, Xesktop can become a flexible creative workspace for anything. 

How to run your own private AI Coding Agent in 2026 

Every time you send code, files, or prompts to a hosted AI service, that data leaves your perimeter. For casual tasks, the convenience is worth the trade-off. But when you’re working with private repositories, sensitive client data, or long-running coding workflows, that convenience becomes a liability.

You lose more than just privacy; you lose control. Proprietary platforms often track usage and traces your work habits. Furthermore, major AI providers reserve the right to use submitted data to tune their models and improve future performance, a standard practice across the industry – take Cursor and ChatGPT for example, their free and low tier plans have data sharing enabled by default. 

If you would rather keep your intellectual property and work habits out of these training sets, you’re in the right place. You don’t need a massive hardware budget to run AI agents locally and privately. 

I’ve developed a streamlined setup to run open-source models using a remote GPU server. No subscriptions, no data sharing, and total control over the environment. Here is how to create your own AI coding agent.

Open-source models vs. Proprietary AI models

Open-source models are closing the capability gap with industry leaders like GPT 5.4, Opus 4.7, and Gemini 3.5, offering competitive reasoning and coding performance. However, the true advantage of the open-source ecosystem lies in three key areas:

Privacy:
By hosting your own models, you ensure that no prompts, code, or work habits leave your infrastructure. Unlike proprietary services, where your data is often utilized to train future iterations of the model, local execution guarantees that your intellectual property remains entirely yours.

Cost Predictability:
While proprietary models rely on subscription fees, local hosting allows for fixed costs via a cloud GPU provider or owned hardware. You are not paying for the convenience of the cloud; you are paying solely for the compute resources you utilize.

Customization:
Open-source models allow you to fine-tune behaviors, system prompts, and tool access without the restrictive guardrails often imposed by closed-source providers. You can optimize the model specifically for your codebase, your coding style, and your specific project workflows.

I’ve tested different models and agents locally and I want to share my setup with you.

1. Software stack

The software we’ll be using is easy to set up. You don’t need to be an experienced dev to get started. 

Inference:

  • LM Studio – a free desktop application to download and serve models on the remote server.
  • Models – Gemma 4 31B, Qwen3.6 27B, NVIDIA Nemotron 3 Super, etc.
  • Hermes Agent – an autonomous agent that lives on your server.
  • Xesktop – GPU inference and hosting platform.

Access:

  • IDE – we will add LM Studio as a model provider in code editor i.e. VS Code or Zed to assist with coding. 
  • Message app – we will setup a messaging channel (Telegram, Slack, Discord) to talk to our Hermes agent in a convenient way. 
  • SSH – we will connect to our remote machine and talk to our agent via SSH. It’s optional.  
  • Agent harness – we can also add LM Studio as a model provider in OpenCode to use the agent locally. 
  • Tailscale – we will setup a private end-to-end connection between our local machine and the remote server. 
Diagram showing a network connection between Xesktop machine and the local machine.

Local machine <–> remote machine connection diagram

2. Server box

For my setup, I’m going to use Xesktop to host the model and run the AI agent on. You can choose any other hosting platform of your choice. Because my machine does not have a good GPU, and I want to run my own model privately, Xesktop makes most sense for this use case. If you have a GPU with 40GB+ of VRAM, you can skip this step and move on to 3. LM Studio setup. 

Server Specs

RTX 6000 Pro Blackwell 96GB VRAM 
32-core AMD EPYC
130GB RAM
500GB SSD

This is quite a beefy machine that allows me to run multiple agents for an extended period of time. Because I will use Hermes Agent on the server with full access to the system, I don’t have to worry about running out of VRAM, RAM or disk space. I will be asking Hermes to install several tools to help me with my day-to-day work and coding.

First, I will create a new system image on Xesktop, give it a name, and select RTX PRO 6000 as my GPU. A few minutes later, once the image is ready, I’ll connect to the machine via RDP (Remote Desktop Protocol).

New system image:

Create a new system image on Xesktop

RDP connection: 

Download RDP file on Xesktop

When the image is created and the machine is launched, we can download the RDP file and then open it to connect to our remote machine. Windows supports RDP natively while on MacOS we need to download the free Windows App by Microsoft in the Apple App store which allows us to import an RDP file. 

3. LM Studio setup

LM Studio is a desktop app that allows us to download and manage open-source models in an easy and intuitive way.  LM Studio has a built-in chat interface so we can talk with the models directly in the app. But what’s more interesting to us in this case is the server feature. We can turn the app into a model server with the http connection enabled for our agent and our local machine to use as an endpoint. 

Let’s download the free app from the official site – we’re going to use the Windows version and install it on the Xesktop machine; no account required.  

Server Settings in LM Studio

Developer tab – Server Settings

Let’s go to the Developer tab in LM Studio, start the server and then under Server Settings enable “Serve on Local Network” and “Enable CORS”. Once they are all enabled, we can load the model and reach it via supported endpoints.

Models

LM Studio supports model weights in GGUF format. There are two ways to get a GGUF model.

A: built-in search 

  1. Open LM Studio
  2. Switch to model search tab
  3. Type a model name (e.g. “Qwen3.6 27B”)
  4. Pick a quantization, click Download 

B: Hugging Face download 

  1. Go to https://huggingface.co
  2. Search for a model + “GGUF” or select LLM Studio app filter https://huggingface.co/models?apps=lmstudio&sort=trending
  3. Look for repos by unsloth, bartowski, lmstudio-community, or similar
  4. Download the .gguf file
  5. In LM Studio: click my models tab → at the bottom open the path → drop the file in

Suggested models

  • Qwen3.6 27B | Q8_0 | 29.5 GB | Code generation, agentic coding
  • Qwen3.6 35B A3B | Q8_0 | 34.4 GB | Best quality/size, agentic coding 
  • Gemma4 31B | Q8_0 | 31 GB | General reasoning, instruction
  • NVIDIA Nemotron 3 Super 120B | Q4_K_M | MoE general reasoning, agentic coding

Search and download models in LM Studio

Choose the model you prefer, download it, and then load it by hitting the “+ Load model” blue button in the Developer tab. To verify the server is running and the model has been loaded run:

PowerShell
Invoke-RestMethod -Uri "http://localhost:1234/v1/models"

then run this command
Invoke-RestMethod -Uri "http://localhost:1234/v1/chat/completions" -Method Post -ContentType "application/json" -Body '{"model": "any", "messages": [{"role": "user", "content": "This is a test"}], "temperature": 0.7}'

The response message should list the model that is currently loaded. 

With that done and the server running, we can now create a private connection.

4. Private connection

Now that we have the model downloaded and the LM Studio server running successfully on the remote machine, we can establish a private connection between two machines using Tailscale.

Tailscale is a zero-configuration free desktop application that allows us to establish an encrypted connection between devices without exposing them to the internet. In our case, we will install Tailscale on both our local machine and the remote machine that is running LM Studio. 

Let’s download Tailscale from the official site. Install it on both machines, then log in to your account to be able to connect both devices. That’s basically it.

Visit the web dashboard to manage your devices: https://login.tailscale.com/admin/machines

In the screenshot below, both of my machines are connected – Xesktop remote machine and the local machine.

Tailscale admin dashboard

Let’s test the connection.

On your local machine, use the same command we used earlier but replace the localhost with the IP address your remote machine has been assigned by Tailscale. 

PowerShell
Invoke-RestMethod -Uri "http://REMOTE.MACHINE.IP:1234/v1/chat/completions" -Method Post -ContentType "application/json" -Body '{"model": "any", "messages": [{"role": "user", "content": "This is a test!"}], "temperature": 0.7}'

Bash
curl -X POST "http://REMOTE.MACHINE.IP:1234/v1/chat/completions" -H "Content-Type: application/json" -d '{"model": "any", "messages": [{"role": "user", "content": "This is a test!"}], "temperature": 0.7}'

You should get a response containing the model info and a reply message. 

5. Install Agent

At this point, we have the model server running and a private connection established. It’s time to install the agent.

I’m going to use Hermes Agent because it’s easy to set up and it’s a well-established project.

Prerequisites
Python 3.11
Node.js v22
Ripgrep
ffmpeg  

Hermes install command
We don’t have to worry about checking and installing dependencies because Hermes will take care of it, including the prerequisites if it detects they’re missing. 

PowerShell
iex (irm https://hermes-agent.nousresearch.com/install.ps1)

Select “Full setup”: 

Hermes Agent configuration

Select “LM Studio” as the model provider and then type a dummy API key when asked to provide one since the server is running locally:

Hermes Agent provider configuration

After that go through the rest of the config:

  • select default model
  • select local for terminal backend
  • select messaging platform (I will use Telegram and will show my setup in the next step)
  • select tools for CLI
  • choose search provider or skip it 

6. Set up messaging channel

I’m going to use Telegram as a chat platform to talk to our agent.  

Hermes has a detailed guide on how to set up a bot on Telegram that we can converse with. Note that setting up a messaging app is just for convenience and it’s not required. We can still chat with the agent inside the terminal cli.

The gist of it is that first we need to create a Telegram bot using an official Telegram bot tool. Then save the access token the tool generates and provide it to Hermes during the initial setup. 

During the interactive config, Hermes will prompt you to use a QR code or a link to setup the bot for you, which is the quickest way, so I recommend doing it that way. 

We can now talk to our agent via Telegram:

Telegram bot

7. Alternative access UIs

We can access Hermes through different UIs. I’ve tried them all and personally like terminal experience, but it’s up to you to choose what’s best for you. I’ll show you how to configure the Hermes Desktop app.

Also, as an alternative we can access the GPU inference directly in our code editor or agent harness such as OpenCode.

Hermes Desktop app

Hermes has its own desktop application, and what’s cool is that we can use it on our local machine while the agent backend and the inference happen on the remote machine.

First, we need to make sure the Hermes web server is running on the remote machine and is accessible from outside since that’s how the desktop app connects to it. The command below launches the web dashboard server. 

PowerShell
hermes dashboard --host 0.0.0.0 --port 9119 --insecure --tui --no-open --skip-build

You can open the dashboard in the browser by going to localhost:9119.

–insecure flag means the dashboard doesn’t force you through the login page. It doesn’t mean the access is insecure because we’ll use session access token instead. 

Let’s create a token by running this command:

PowerShell
python -c "import secrets,base64; 
print(base64.b64encode(secrets.token_bytes(32)).decode())"

Bash
openssl rand -base64 32

We need to save the token in C:\Users\USERPROFILE\AppData\Local\hermes\.env as:
HERMES_DASHBOARD_SESSION_TOKEN=<your token her>

Now back to our local machine, let’s install Hermes Desktop and then launch it.

We can download the installer from the website or run this command to install it just like we did earlier on the remote machine. It’s the same Hermes Agent. The only difference is that here we’ll launch the desktop app as our primary UI.

PowerShell
irm https://hermes-agent.nousresearch.com/install.ps1 | iex

Bash
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

Let’s start the desktop app by running `hermes desktop` in the command line.

Then go to Settings > Gateway and paste your remote server URL and the session token.

Set up remote gateway in Hermes Desktop app

Gateway settings in Hermes Desktop – remote server URL “http://REMOTE.SERVER.IP:9119”

After saving and reconnecting, we’ll be able to use the desktop app with the agent installed on the remote machine.

Zed (IDE)

In the IDE of your choice, you should be able to add a custom OpenAI API compatible LLM Provider. In my favorite IDE – Zed, I added our LM Studio as a provider.

We need API URL (this is the same URL we have used earlier to test the LM Studio server connection: http://REMOTE.SERVER.IP:1234/v1), API key (i.e. “dummy-key”), and the models you have downloaded and want to use (i.e. qwen/qwen3.6-27b, google/gemma-4-31b etc.).

Add a custom LLM model provider in Zed

Add a custom model provider in Zed

OpenCode

Adding a custom provider to OpenCode is a breeze. Let’s locate the config file and then add LM Studio remote.

Path to config: .config\opencode\opencode.jsonc

OpenCode custom model provider

After that, let’s restart OpenCode and user the “/models” command to select our newly added provider. We can now use our remote model locally to code.

Switch a model in OpenCode

Final thoughts

You’ve successfully built a fully private AI coding environment. You’ve bypassed the constraints of proprietary platforms and kept your data where it belongs – with you.

While this setup is a powerful starting point, it’s also a good place for experimentation. I encourage you to swap in new models, test different agentic workflows, and tweak the setup as your needs evolve. The landscape of local AI moves fast, and you now have the foundation to stay at the cutting edge without ever having to compromise your privacy.
Use this setup to push your projects further and keep exploring!

Resources

—
Lucas

GPU rental and other services: How the digital world can help you with a deadline

The year 2020 has been momentous if anything. In the world of 3d, we’ve seen some developments underway that promise faster workflows, more accessible software, and new ways of interacting with digital media in three-dimensional space. We’re also challenged more than ever to work flexibly and produce quality content within shorter time frames, possibly outside the physical proximity of our teammates and collaborators.

Now more than ever would be a good time to take advantage of the many resources online- from GPU Server rentals, industry level open-source software, render farms, and more- to help us land more jobs, or get our digital content ready as soon as possible.

As 3d artists and designers in the CG industry, we may soon find ourselves amidst growing competition, and in more project-oriented work than a stable monthly income. Investments in hardware may be less feasible for some time, and we may need to land more projects to get by. For some, now might be the time to focus some energy on original content. In any case, we are looking at a new decade of uncertainty, but we face it armed with all the resources our digital world has to offer.

Third-party services

GPU server rental

Xesktop

Xesktop is a remote GPU server rental solution that provides access to high powered GPU rental servers for $6/ hour. This can come in handy for offloading heavy computational tasks from a personal workstation, and as an alternative for GPU rendering when a render farm isn’t justified. Users may choose between servers equipped with 10 GTX 1080 Ti Cards with 11GB vRAM or 8 Tesla V100 cards with 16GB vRAM. Xesktop also offers 24/7 support and a free hour for trial and set-up.

A workstation set up is stored as a virtual image so that users can return to their existing projects at any time.

Render Farm

render farm server room

GarageFarm.NET

GarageFarm.NET is a render farm that offers competitive pricing and 24/7 Live Support for its users, as well as a proprietary scene preparation plugin for popular 3d software and Render Engines. They have also recently announced new support for GPU rendering on their farm. The difference between GPU rental and using a render farm is that with the former, a user has direct access to a GPU workstation and can work on projects there. The latter is meant strictly for rendering. With GarageFarm.NET, new users get $25 worth of starting credits, and often give coupons and promotions that can double the starting amount. They are also open to subsidizing a considerable amount of rendering in exchange for participation in case studies and project showcases as well as other content contributions.

For large scale projects, they give huge discounts for large top-ups. Refer to their pricing page for more information.

Software

3d dccs can be expensive, and upgrading to the latest release, or maintaining subscription might have to be postponed in light of recent events. Fortunately, viable open-source or freeware alternatives are available as partial or total more sustainable alternatives for tools crucial to our pipelines.

Blender

Blender render software dashboard

Blender is an open-source 3D creation suite. It has grown to be a formidable tool for every part of a 3d production pipeline: modeling, rigging, animation, simulation, rendering, compositing, motion tracking, and video editing. Despite being a bit of an underdog in its earlier years, Blender is now recognized as a viable 3d package for industry-level use. Not only is it used by notable studios, but it has also garnered the interest and support by way of a financial grant from many industry-leading companies such as Ubisoft, Epic Games, and Nvidia.

Quixel Mixer

Quixel Mixer logo

Quixel Mixer (previously known as Megascans Studio) is a tool that blends scanned surfaces together to create tileable texture-sets for Film and Games. The recent 2020 updates provide features akin to Substance Painter’s fill layer system, where more control is afforded users by way of paintable masks and mesh maps. Mixer is also available for free, without restriction.

Unreal Engine

Unreal Engine rendered scene

Unreal engine is a 3d creation tool and game engine centered on game development and real-time rendering. The software is also made available for free by Epic games, under very liberal terms of use. Many industries turn towards the advantages of real-time rendering in terms of speed and immediate visual feedback during production and Unreal’s continued development in realtime rendering promises closer levels of realism to traditional rendering with every release.

Since UE requires some serious GPU power and no straightforward way of distributing the render process across a network, GPU rental services can come in handy for those without dedicated GPUs looking to render animated sequences from directly within the engine.

Photopea / Krita / Gimp

Image manipulation software is integral to 3d creation as an auxiliary means of texture map creation and editing, and postprocessing. Luckily, there are a few capable tools available under open-source licenses or similar.

Photopea

Photopea software dashboard

Photopea is a browser-based software patterned tightly after Photoshop’s interface and toolsets. The only disadvantage being a lack of hotkey support, Photopea is a great alternative to Photoshop and offers little to no learning curve for transitioning Photoshop users.

Krita

Krita rendered scene
Krita software dashboard

Krita is an open-source digital painting and texture creation program from the Krita Foundation. While its user base leans heavily toward concept art and 2d painting, its texture tools allow an intuitive way to create and test tileable patterns. Krita also has a thriving community where many useful brush sets and other tools are contributed by users for free or at very affordable prices.

Gimp

Gimp logo

GIMP is an image editor used for image manipulation, drawing, and processing tasks associated with Photoshop, and is available for free under the open source license. While it’s selection tools may not be as refined, Gimp is certainly a capable tool for mask generation and texture map editing.

Assets

Models, textures, volumes, and hdri maps are crucial to most projects but can be costly. Unfortunately without pre-existing assets, projects can take twice as long or more to finish, which could mean the inability to meet deadlines. Fortunately, there are several places these assets can be found at low cost or even for free.

Textures.Com

Textures.com is one of the oldest and most extensive repositories for tileable textures for 3d, as well as mattes and recently, 3d assets. While the platform is subscription-based, a free account grants 15 credits every month that may be used for many of the assets available.

3d model haven

Texture haven

Hdri haven

These three platforms are repositories for models, textures, and hdri maps respectively, and are donation based. The sites are run by open source advocates, Greg Zaal, Rob Tuytel, and Cameron Casey. All the assets are available to download for free with a CC-0 license, which means they can be used for any purpose without the need for accreditation. The sites are sustained by the community through support on Patreon, and grants from companies like Epic Games.

3DBee.IT

3DBee.it website

For Interior Architectural visualization, 3DBee.IT offers a growing library of high-quality 3d models, materials, and optimized scans for furnishing, appliances, ornamentation, and food. While the service is subscription-based, affordable credit packs are available for on-demand purchases, and many of their assets are available for as little as $5 or free.

Summary

While these resources are just as useful under normal circumstances, the huge advantages of leveraging services and platforms such as these are the time saved in content production and the flexibility in allocating budgets towards what is most profitable in any given situation.

With the state of many industries unpredictable in times like these, it’s a comfort to know that many alternatives exist, and are a google search away.

5 Situations in 3D Production Where GPU Power is a Must-Have

By Danny Rollings | @drollingscaa

It would be wrong to cover the power of GPU without explaining just what it is. Initially, the meaning of this acronym can be a bit of a mystery, conjuring up images of complicated machinery, unless of course your knowledge set is that way inclined. GPU stands for Graphics Processing Unit, as opposed to CPU, or Central Processing Unit (more on that later). If I said the words graphics card this would provide an abundance of clarity to any dedicated PC gamer – of which I’m not, PC “master race” be damned. No, I was unfortunately led to believe Mac was far superior for 3d animation, and near shed a tear when a friend said they’re gonna get Borderlands 3 on PC. Technically a graphics card is not the GPU, but rather may include one or more GPUs as part of it; although visualising those semi-futuristic looking rectangular things with the dual (or even triple!) fans gives you a good idea of what we’re getting at.

Composite image of the AMD Radeon VII
Composite image of the AMD Radeon VII, “Engineered for Enthusiasts”
(source: AMD)

While it’s true an individual CPU core of a computer or games console is good at performing a large variety of tasks sequentially very fast (making me a nice cup of tea is sadly not one of them – yet) and a single core of a GPU is slower and simpler than those of a CPU, this is made up for by the sheer volume of GPU cores and their ability to work in parallel. For example a top-notch 18-core Intel i9-7980XECPU will set you back at least £1,785 (US$1,843), meanwhile the 4352-core self-described “ultimate gaming GPU”, Nvidia GeForce RTX 2080 Ti costs £1,099 ($1,199); ⅓ cheaper for over 241× the cores. This quantitative approach becomes all the more valuable when you consider the technological limits of individual core improvement, you can only do so much.

difference between CPU and GPU

1. Simulations

Simulation of two spiraling supermassive black holes about to merge
Simulation of two spiraling supermassive black holes about to merge.
(NASA Goddard, 2018)

With the word simulation soon comes the idea that we are living in one; we have The Matrix (1999) to thank for that. We’re not talking that kind of simulation, but the at-times-beautiful visualisations hiding mathematical processes that range from relatively simple to the mind bogglingly complex. The most famed simulations replicate interactions of celestial objects such as stars, galaxies, or black holes. They allow us to speed up the colossal timescales of the universe and view sections of it how they once were or what they’re going to become via supercomputers. Of course, you don’t need a supercomputer to perform a simulation of your own, after all most 3d software comes with the ability built-in to some degree.

Let the Avatar: The Last Airbender binge-watch begin
Let the Avatar: The Last Airbender binge-watch begin!

By messing around with the properties and interactions of a bunch of floating balls or dots you can simulate all four of the classical elements and beyond. If you want to create as high quality a simulation as reasonably possible, you’re gonna need a lot of power and some damn good software. This is where plugins such as Phoenix FD for Maya & 3ds Max (£40 per month) or Trapcode Form for After Effects ($199) come in.

A variety of Phoenix FD 3.0 for Maya quick presets
A variety of Phoenix FD 3.0 for Maya quick presets (Chaos Group, 2017)

If you want your simulation to look not only great but render as fast as possible, you’re gonna need GPU power. Don’t just take my word for it; a 2017 paper by members of the NASA Ames Research Center, NIO & the USRA states “The GPU-based implementation provides a significant advantage, and scales much better than the purely CPU-based implementation. With a small number of samples…the CPU-only approach is preferred. For larger numbers of samples, the GPU approach provides an order of magnitude improvement.” Think of the GPU as an army of master swordsmen, whereas the CPU is a trained assassin capable of dispatching foes one by one with a variety of weapons.

GPU acceleration prognostics graph and table
Source: GPU Accelerated Prognostics, pg. 5 (colourised)

2. Real-time rendering for previsualisation

Pixar’s Monsters, Inc. (2001) animation pipeline
Pixar’s Monsters, Inc. (2001) animation pipeline (source: YouTube)
Animatic Storyboard > Previz > Animation > Render

With the power devouring waiting game that rendering usually is, real time rendering may sound like nothing more than a pipedream. But it’s actually something that many of us see regularly and take entirely for granted. Real-time rendering is central to today’s videogame industry, and it relies heavily on the GPU power. The visual splendor of today’s games demand it. Everything from the ever-changing XYZ position of your character, to the lighting, textures, and various other simulations such as fire and gravity need to be re-rendered simultaneously time and time again within fractions of a second. As to real-time previz, you may be surprised to know it has been around since at least 2006, when it was first developed by ILM.

War for the Planet of the Apes (2017) bts footage
When motion capture and real-time previz come together apemazing things can happen.
(War for the Planet of the Apes (2017) bts footage. Source: YouTube)

Now here’s where realtime pre-viz comes in. Because of the nature of animation production how closely the previz resembles the final product depends on time constraints, budget and team size. While movie studios can afford to create action packed & easily interpretable previz, the same can’t necessarily be said for a potentially struggling freelance 3d artist.

Being able to see a higher quality version of your previz in the viewport of one window, while you’re making adjustments in the other is a game changer for not only productivity, but clarity too. After all, the better your previz, the more chance you have of an interested party getting on board with your project.

3. Traditional rendering

The benefits of GPU rendering are similar to some of the benefits of using a render farm. First and perhaps foremost it saves time. After all, handling complicated graphics processing is the name of the GPU game. As Workstation Specialist helpfully points out, GPUs are designed “to render on specific render software packages available in the market today such as, NVIDIA’s IRay, Chaos Group’s VRay RT, Otoy’s OctaneRender; and Maxwell Render” which is comforting to know if you’re considering buying a new graphics card for rendering purposes.

A comparison of 2004’s DOOM 3 and 2015’s Battlefield Hardline hamburgers
For those hungry for useless knowledge, pictured are 2004’s DOOM 3 and 2015’s Battlefield Hardline.
(Source: 1 2)

So while your computer’s CPU is straining to keep your web browser from collapsing under the weight of far too many tabs (guilty as charged), your GPU doesn’t give a damn and renders those frames at comparatively lightning speed; however if you want to do a binge-watch of a series or videos while you wait, you’re better off using another device. You could also try the very interesting option that is hybrid rendering. If this isn’t possible, or you are looking to be as cost effective as possible, you may want to consider using a render farm instead, which, as paradoxical as it may sound, is a far more cost-effective solution to cashing out on a brand spanking new graphics card or CPU.

4. Manipulating heavy scenes

Particle simulations partly fit into this category, but this depends on what you’re going for; this could be a simple spilling glass of milk, or say, the usual colossal tsunami destroying Hollywood’s favourite (or least favourite??) city. As 3d designers we have become so incredibly sensitive to viewport lag. To avoid our chosen software crashing we waste time carefully watching our polygon counts and “fake” as much of the complex geometry as possible via texture maps, spending hours upon hours dutifully plugging them in and hoping for the best, just so our computer doesn’t melt. Just imagine sculpting something incredible in Mudbox and being able to render it as is, with GPU power that dream can be a reality.

Many memes on Maya
Many a meme have lived and died over the years, but one thing has remained the same…

5. Texture painting

For this one we’ll be focusing on the award-winning Substance Painter by Allegorithmic (no affiliation); who were acquired by Adobe earlier this year. This software streamlines the massive undertaking that is texturing, rendering and effects application via a physically-based rendering (PBR) workflow. Just what is all this exactly? Well, SpeedTutor has got you covered.

With SP’s high importance to the likes of AAA game developers, it’s clear why many in 3D production would want to get in on it. Painting textures directly onto your model is a far cry from the dull and repetitive process that is UV unwrapping. It’s a process that is a beautiful emulation of reality, the pride and joy of any digital Banksy. This is all the more tempting by its 4 license tiers. The Indie license alone covers any revenue below $100K (~£77.3K) for as little as $19.90/mth or $239/yr (~£15.40/mth, ~£185/yr), and let’s not forget the discounted Education license. If you’d like to see if your computer can survive, you can download a 30 day free trial. Unfortunately it will be of no surprise to hear that SP is extremely GPU intensive. Not being able to run this arguably beautiful software can be absolutely heartbreaking for any freelancer, from newbies to veterans; but don’t worry, we’ve got your back.

By now there should be little doubt in your mind of the immense benefits of GPU powered 3d production. But don’t worry if you can’t afford to invest in a whole new graphics card, or you’re in the minority of Mac users seemingly forever stuck with a far past its prime GPU—such as my defunct 2013 NVIDIA GeForce GT 750M—and can’t afford an eGPU, GarageFarm.NET have created an alternative that is both cheap and powerful, a GPU-server rental service that goes by the name of Xesktop, at your disposal for GPU 3d rendering, processing Big Data, or any task that can benefit from parallel processing.

Xesktop server specs