Render Network AMA / Q3 2024
Hey Render Community,
It has been an exciting start to the year. AI and GPU are the hottest acronyms on the lips of every tech startup we can think of. The Render Network is the first decentralized GPU network to cater to the needs of 3D artists by providing a near limitless number of GPUs to accelerate rendering workflows from days and weeks to minutes and hours. The flexibility of the Render Network allows us to also tap into the emerging sector of AI, from compute clients requiring more GPU power to the emerging sector of Generative AI.
Over the past few weeks we have collated some of the key questions from the community. With such a wide range of questions — from those getting started with creating art to seasoned veterans patiently waiting for key updates — we decided to split this AMA in two parts; an Intro and a Technical AMA.
Let’s dive in!
Render Network Intro AMA:
1. Where do we find information about the foundation?
2. What’s the purpose of the foundation?
We are a not-for-profit dedicated to maintaining the core Render Network protocol and growing its community.
3. What render engines are supported except Octane?
The main engine supported is Octane, as it scales very linearly and has support for pretty much every digital content creation app — so everyone using Octane can effectively use the render network.
Other engines are currently in alpha/beta and are due for release after further testing, bug fixing, and development etc. These are:
- Redshift (works but users need to be manually added to the beta)
- Blender’s Cycles
- You can also upload C4D files directly using the native Standard, Physical renderer, or Octane.
4. What file formats are supported except orbx?
ORBX export is a must on all 3D apps for the moment. C4D users can upload .c4d files directly (our first native integration), however we are working on making that as smooth as possible and are working on a Wizard which will help with the process. This Wizard should also be useful for developing a similar tool for Blender once we have Cycles on the network.
5. How can a PC owner add his/her own PC to the render network?
Users can apply at renderfoundation.com/gpu — there is a long waitlist but we’ll be adding more of those that sign up as usage of the network increases.
6. Is there a dedicated website for artists with simplified information about the service and it’s costs?
Currently we have our Knowledge Base, along with a few artists that have produced How-To videos. We are also working on creating more content for users.
https://know.rendernetwork.com/getting.../how-to-get-started
Here’s Brilly showing how to get started:
Render Network Technical AMA
Q1. Are we getting close to completing integrations with additional render engines? Is there something that needs to be implemented before these integrations can go live, or has other work been prioritized ahead of the integrations? Do you think having more developers solely focused on the Render Network would help expedite these integrations?
Multi-Render integrations started with Arnold Render, but the near-term focus has shifted to Maxon’s Redshift and Blender’s Cycles. These integrations present the greatest opportunities and overlap with current users. Significant progress has been made on both fronts. Today, Maxon’s Physical and Standard 3rd-party CPU renderers are integrated.
Redshift is in alpha testing with promising results so far. We are working with the Maxon team to extend and refine Redshift to resolve blocking issues currently preventing us from moving to beta. This is a complex process given the decentralized architecture of the network.
We have also started integrating Blender and Cycles, showing good progress due to our experience with Redshift. We are not yet at the alpha stage but are moving in that direction with a roadmap for implementing a similar ‘Wizard’ application to make Cycles usage more frictionless on the Render Network.
The Arnold integration announced four years ago was very different from the deep C4D+ Redshift and Blender + Cycles integrations in process. The concept is to make Standard Surface a core material system standard that would work natively on the Render Network, enabling users to mix and match renderers like Arnold and eventually Brigade. Starting with Redshift and Cycles as core integrations, Arnold and UE are potential candidates for this type of deep integration if the community prioritizes.
It is not simply a matter of adding more developers; these are complex dependent systems that involve third parties.
In addition, it is important to remember that the network is not simply a distributed render farm, which actually would be much simpler to implement. The Render Network builds a layer of granular transparency in terms of provenance of art & 3D assets, which will provide future value to artists.
Q2. Has the team started marketing/pushing the use of the network to the other 99% of Octane users?
Currently, our efforts are more about pull than push, focusing more on awareness than pushing use of the network per se. This is because it currently requires a more expert-level user to get the most value from the network. The team is prioritizing making the network easier to use for the average user before broader outreach and has made significant progress. The team is in the process of completing the C4D plugin Wizard (RNP12) that simplifies moving scenes from C4D to Render and includes automated checks and fixes for common issues observed over the past four years.
The rollout of the Wizard will begin more widespread outreach and will include a prominent link to the Render Network from within the C4D Octane plugin, representing the start of a broad marketing push across C4D + Octane and C4D + Redshift users (once Redshift is unblocked by Maxon).
This will be paired with emissions set aside for creator rewards via several marketing initiatives, and the completion of the BME implementation with simplified RENDER artist-side purchasing. With these in place, the network will be better positioned for large-scale outreach, including prominently featuring Render during the largest Octane event of the year, the annual Black Friday Octane release.
In user outreach, an important factor to consider is attrition and conversion rates, with word of mouth within the tight-knit ecosystem of professional 3D artists adding to these. For now, we are focusing on catering to and learning from lead users, with the goal of being well-positioned to achieve high conversion rates on wide outreach, and we are making great progress.
Keep an eye on our socials — when we start onboarding more artists you’ll definitely notice it.
Q3. Realistically, how long do you think it will take to have the Stability AI models optimized and running on the network?
Stability AI models are currently run on the network via Compute Clients (e.g., ionet’s BC8 runs inference on a fine-tuned stability OSS model). These are not yet utilizing the BME model, and io.net is still completing their integration.
If the question relates to Emad’s RNP011, it’s honestly too early to tell as we need more information from Emad, which is one of the reasons why we have not pushed the RNP forward yet.
Q4. Are there any significant updates that can be shared regarding dynamic NFTs that the team was discussing with Beeple, $M for Metaverse, and/or the artist/creator monetization opportunities that have been discussed previously?
Nothing to share here at the moment. We have had active discussions regarding both recently and hope to have more to share as these discussions continue.
Q5. For the last two years or so, messaging has been that the team is focused on increasing network usage. Have those efforts been successful? What other approaches has the team considered?
Yes, usage has consistently increased every year since the network was formed.
There were several key learnings over this period that led to the integration of native C4D files with Redshift, a C4D wizard, Blender + Cycles, and allocating a significant percentage of emissions to artists (which started this quarter).
Q6. When will the API/SDK be open to the public?
Both Compute Client and Render APIs are available for integration, and currently have third parties working on integrations. There is no timeline on when we would make these publicly available, but it’s not hard for anyone to approach us, and if their project makes sense, gain access via the Render Network Foundation and Grants program.
Q7. Does the team think some of the functionality of the compute clients will be added directly to the network? For example, wouldn’t it benefit the Render Network to receive AI/compute workflows directly instead of through a compute client with its own token economy?
Perhaps this is more a question for the community as it would require significant engagement.
Some of the key reasons we supported an API approach include:
- As of yet, none of these compute clients have achieved hypergrowth (product market fit) using decentralized consumer GPUs. Should that change, it would provide a good reason for the community to consider a more direct approach.
- Compute clients who use render nodes drive value to the Render community. Currently, most of the Compute Clients are actively chasing their own scarce, high-end Enterprise GPUs, whereas we are focused on creators who own an abundance of high-end consumer GPUs.
- There are a lot of exciting AI applications building off popular image generators like Stable Diffusion that can run on consumer GPUs, and we think these would be the best fit for native integrations, e.g., apps building their infrastructure directly on the Render Network.
Q8. With the opt-in option for serving compute clients as a node operator and the initial writeup of RNP-011, it seems the node structure is becoming more fragmented and confusing with several layers and tiers of nodes. What are the benefits of having a somewhat fragmented node pool with some nodes being core render nodes, some nodes only serving compute clients, and others serving both?
Running separate pools per client has allowed us to limit our investment while these clients find product market fit.
The architecture for Compute Clients is very different from rendering. Each client needs a specific Docker/Kubernetes VM pre-installed along with their node software running.
Render nodes run a very thin client with the software and job streamed to them to maintain security. It will be a complex task to merge nodes.
Q9. What metric is used to determine when to add a node from the waitlist?
Utilization. We need to be closer to 10X where we are today. This is a moving target as existing nodes keep upgrading and therefore increasing the overall network OB.
Q10. What are the largest hurdles/challenges that you think the Render Network faces over the next 1–2 years?
We are in the middle of several disruptive changes:
- Open-source DCCs starting to gain traction vs. paid
- AI disrupting DCCs
- AI/neural rendering/diffusion emerging as methods for creating content
- The transition of the internet to 3D
All will need GPU power, but it will be critical to remain nimble and adapt our offering to maximize the opportunities that emerge from this.
Q11. What do you think has been the biggest missed opportunity (if any) for the Render Network over the last few years?
That’s hard to say, as the network has done really well in positioning. For example, we acted early on the Solana integration, being one of the first large networks to transition L1s, and that has turned out exceptionally well. The network also integrated AI into its roadmap in an actionable and thoughtful way, focusing on 3D AI models with DRM tools that appeal to, rather than compete with, creators and media and entertainment organizations at large.
Possibly not being able to focus on open-source DCCs earlier was a missed opportunity, but adding Blender Cycles to the network as a deep integration is a huge step forward in that regard. We are very excited about it.
Q12. In hindsight, given the current burn numbers and the recent RNP to reduce node rewards per epoch, do you think you would’ve delayed the implementation of BME until network usage increased?
If we had delayed, we would not have had the benefit of being able to allocate emissions to areas like creator rewards. My understanding from Helium is that they regularly rebalance, and a lot changes in six months.
Q13. Is there any concern about the lackluster voter participation in RNPs? Has the team thought of any ways to further promote or incentivize participation in network governance?
It is frustrating. Holders on CEXs can’t vote, and larger holders have genuine legal concerns. It is not an incentive issue; hopefully, someone in the crypto ecosystem comes up with a better solution for all, as this is a crypto problem, not a Render Network problem.
Q14. What are the top 3 things the community can do to help the Render Network?
Participate, vote, and market the project.
We are increasing our social network presence on a wide range of social platforms — we encourage you to engage with our content!
Q15. What steps are being taken to lower the barriers for integrating into the Render Network? Current RNPs like Ionet and Prime Intellect require “integration” and “plugins.” Are we moving towards a future where individuals can easily use the API to leverage the network on their own?
It’s not that simple. If we wanted to, we could give key access to nodes via API, but there are obvious trade-offs. Creators don’t necessarily want to give up root access and need to use their nodes periodically.
Otherwise, some software needs to be pre-installed (which is what we are doing for each compute client pool).
I would challenge the person asking this question to come up with a use case.
Q16. When will it be possible to use macOS or iOS as a node (iPad or MacBook)?
This is dependent on several factors:
- For render nodes, Apple devices need to achieve sufficient rendering OB (they are improving with each new M series release).
- We would also need increased usage to the point of needing to open the waitlist.
- For Compute Clients, this could be sooner, as io.net has made some progress getting Ray to run on macOS.
Q17. What do you anticipate being the greatest driver of demand for the network in the next 6 months (i.e., calendar year 2024)? Can the team be more transparent with timelines and KPIs? Given that Jules’ focus is multi-decade, it’s very difficult to understand the timeframe on certain initiatives. More frequent opportunities for AMAs with the community could help a lot of FUD that grows only because there isn’t enough communication.
It’s hard to say which will win out of all of these:
- C4D wizard
- Redshift official release and marketing
- Creator rewards
- 3D, Vision Pro, and large-scale sphere-type demand
- Expanded VFX usage
- Blender and Cycles
- Third-party render API integrations
- Existing compute client PMF
- New compute clients
- Black Friday release with tighter Render integration
Almost all have third-party dependencies, which is why timelines are not preferable. The best place to track progress will be stats.renderfoundation.com and X.
Q18. Jules once mentioned that the plan is to support Hugging Face spaces as modules. What is the status of this?
Nothing to share here at the moment, but stay tuned.
Q19. Emad is talking about some new decentralized AI project. How will Render be involved in this? Exciting!
Yes, it is exciting, but it’s still too early to tell. We are looking to Emad to provide more guidance here.
Q20. A compute client integrated with a competing crypto project before Render, even though they were announced as a compute client first. Is there anything holding clients back from integration, and will there be KPIs and deadlines for compute clients going forward?
Compute Clients, of course, have integration deadlines and performance milestones. We are continually learning from successive engagements, and our processes are always being refined. It is in our interest, as well as our compute clients, to find a product market fit and to have usage increase.
We hope you enjoyed this AMA update and found the answers enlightening. To stay informed about all the latest news, updates, and events, be sure to join our social channels. Connect with us on X, Telegram, Discord, Facebook, Reddit, and LinkedIn, to keep the conversation going and be part of our vibrant community.
Thank you for your continued support and enthusiasm for the Render Network. Together, we are shaping the future of decentralized rendering and AI technology.
Make sure to join our weekly Spaces on X, and in the meantime, tune into today’s session:
Stay tuned and stay connected!
The Render Network Foundation
