Announcing Pixel Plow

[Cyber-Angel]

A few questions:

1: What is the total node size of your farm in comparison to say well known farms like the Death Star Farm at ILM and the one at Weta Digital?

2: Is your farm capable of handling feature level work with high levels of simulation work e.g 2012, Perfect Storm?

3. Related to question 2! How much storage dose your farm have is it in the GB, TB or PB range?

4. Roughly what would it cost to submit a 2hr feature animation at 129600 frames at 4k cost?

5. Do you have plans for Houdini and Trapcode Plugin Support in future?

Regards to you

Cyber-Angel           

[digitalis99]

A few questions:

1: What is the total node size of your farm in comparison to say well known farms like the Death Star Farm at ILM and the one at Weta Digital?

2: Is your farm capable of handling feature level work with high levels of simulation work e.g 2012, Perfect Storm?

3. Related to question 2! How much storage dose your farm have is it in the GB, TB or PB range?

4. Roughly what would it cost to submit a 2hr feature animation at 129600 frames at 4k cost?

5. Do you have plans for Houdini and Trapcode Plugin Support in future?

Regards to you

Cyber-Angel         

1) For the time being, details on what constitutes our farm are remaining proprietary.  We have too many unique functions in facilities, hardware, and software to be directly comparable to existing farms.  That uniqueness is also what makes us better than other farms, hence our desire to maintain information privacy.

2) Absolutely.

3) Due to our operating model, we don't require nearly as much storage as any other farm.  We deliver results as they are created and purge them automatically after they are confirmed delivered to our customers.  We're not an online backup provider like other farms become, so our storage needs are significantly reduced.  Like #1, though, this information remains proprietary for the time being.

4) I can't even begin to answer that question without a lot more information.  There are far too many variables that would have an absolutely massive impact on total compute time, and hence total cost.  Are you planning on producing such a project now?  PM me if you are.

5) Yes, Houdini and Cinema4D are the next two apps to support, which we hope to have in place within a month or two.  Plug-in support is going to depend quite a bit on popularity and user demand, so if you're planning on making heavy use of a particular plug-in in conjunction with our service, please let us know.

Thanks!

[Cyber-Angel]

Thank you for your reply I am thinking about a project, some time in future (Don't have a definitive time-frame yet as it is still in the planing phase) I was just sounding you out to see what might be possible down the line: what with needing to submit test renders and simulation tests and all.

You sound more cost effective than going to a large production house (Not that I'd know how you'd go about that nor whom to talk with about such an enterprise).

Thank you again.

Cyber-Angel 

[TheBadger]

TY
I have not used a render farm yet. So I am sure I will bother you when my job is ready for sending.
I have been working on and off on these projects for a while now. And I have every intention of having them done in the spring. But I really need to plan in advance when thinking about paying for rendering.

[digitalis99]

TY
I have not used a render farm yet. So I am sure I will bother you when my job is ready for sending.
I have been working on and off on these projects for a while now. And I have every intention of having them done in the spring. But I really need to plan in advance when thinking about paying for rendering.

Proper planning is always a good idea.  Good luck on the projects!

[digitalis99]

... You sound more cost effective than going to a large production house...

It's funny you mention this, because we're attracting attention from some large production houses as I type.  Apparently, they are having a hard time competing with our pricing using their own internal farms.  That says quite a bit, I think.

[digitalis99]

Just a little update on new features:

1) We support and use the naming convention defined in your .tgd's now as regards output file names, both primary and additional.  This is very handy for people re-running the same scene multiple times when the outputs are used in compositing applications.
2) We have fully automated budget handling now.  Previously, we would only suspend jobs in an automated fashion if the budget estimate exceeded the user-defined amount.  Resuming the job required admin intervention on our side.  Now, each user can resume jobs directly from the queue manager in our software.  Our over budget alert emails (which are sent the instant we estimate your job will cost more than your defined budget) also include the estimated project completion costs.  That enables you to make the decision as to whether or not you want to continue the render beyond your budgeted amount, rather than having to guess at how much the total job will cost.
3) Client-side queue management now works correctly on Windows 8.  Jobs were always handled correctly, but display of jobs and job status on Win8 was unreliable.
4) Minor GUI additions to make job tracking and status clearer and more informative.

All customers already have the updates available on their machines.

[FrankB]

This is great. I might have small job to submit soon!
Like Dune, I am not excited about credit card being the only payment option though.
However I understand your reasoning. Perhaps it would be a bit soothing if the CC details would be deleted after the transaction.
Anyway, looking forward to trying you guys out!

Frank

[digitalis99]

This is great. I might have small job to submit soon!
Like Dune, I am not excited about credit card being the only payment option though.
However I understand your reasoning. Perhaps it would be a bit soothing if the CC details would be deleted after the transaction.
Anyway, looking forward to trying you guys out!

Great, we'd love to have you on board, Frank!

Deleting the CC information (which we can do if you're adamant about it) would require you to re-enter it for every job, at which point we'd have to re-validate it to make sure it's chargeable.  I'm not aware of technology that exists to enable us to *not* have some sort of stored and chargeable payment method...other than massive pre-payment, of course.   

Credit cards are really the only fair way to do this.  If better options present themselves, we'll likely implement them.

[gregtee]

I just want to chime in that Pixelplow rocks. They're very inexpensive compared to other services in this area, they allow multiple versions of TG to render on, their submission process is clean and easy and you get your frames as soon as they drop off the farm.

I've run thousands of frames through their farm over the last couple of weeks, from quick test renders to full 2k passes, all without a hitch.

Highly recommend.

-Greg

[digitalis99]

Thanks gregtee!  If this forum had a "Like" button, I'd have clicked it.

We've added more features recently.  A few of these features work in conjunction with each other to function more effectively.  The straight list is:

1) We support frame lists rather than just a single start and end frame (eg. 1,5,17,34,200:367)
2) We render frames and tiles in random order by default.  The old method of sequential is available as an option, but random makes all of the cost/budget features much more accurate earlier in the job run.
3) We support a per-job output folder now
4) Budget Goals are available at all priority levels
5) We support changing priority up or down at any time during the job run
6) We now have a running total job cost estimate displayed in the queue manager
7) We deliver frames to the default location if the per-job location is unavailable at time of frame delivery

There are more non-TG features and enhancements, but I won't go into those here.

We're really excited about #6, since it displays a total job cost estimate while your scene is rendering.  This feature is independent of the Budget Goal setting, which is a nice automated way to suspend your job if its cost starts to run away.  In order to make both money-related features work as effectively as possible, we implemented #2.  Most scenes, especially TG scenes, were too easy/quick to render in the beginning of the frame range.  That delayed the accuracy of the Budget Goal and Job Cost Estimate features until late in the scene runtime on the farm.  Now that we render in random order by default, both estimation-based tools will be closer to reality earlier in the process.

The Job Cost Estimate happens automatically for all jobs, at all priority levels, whether a Budget Goal is specified or not.  The Budget Goal is basically just a fully-automated way of having our system "watch" the Job Cost Estimate and immediately suspend your job if that estimate exceeds the Budget Goal you defined.  We've given you the ability to override your budget and resume rendering from our queue manager interface, so nothing on our side requires human interaction or involvement.

The frame list feature was requested by multiple users, so that is another addition we're happy about.  As well, the per-job output folder is pretty handy if you're doing compositing work on the rendered frames.  Between that and honoring the file naming output defined in your TG scene file, successive runs of the same scene can deliver their output to the same location with the same naming scheme.  It's an addition that's very helpful in larger workflow environments.

[digitalis99]

Just a quick update:  We added display of the percentage of completion for tiled stills in the queue manager.  Now you can see the job progress in real-time whether it's a tiled still or a frame range.  This can help you gauge both an ETA for tiled jobs as well as determine whether or not to change job priority.

Non-TG sidenote: If you want us to support Houdini, contact your SideFX rep and tell them to remove their backwards EULA restriction on public render farms for us.  We're working with their higher-ups, but they need some persuasion.

Non-TG sidenote 2: C4D support imminent.

[digitalis99]

There have been many non-TG updates and application support added since my last post, but there's a more important bit of news to convey.

Over the past few months, we've purchased and built-out our own facility to house our farm.  This new facility gives us access to copious amounts of inexpensive electricity, year-round nearly free cooling, lots of physical space, and more.  Basically, we've moved out of the lab and into the facility that enables us to quickly grow to meet customer demands.  We've gone big time.

In addition, we've worked on a spreadsheet that enables studios to easily boil the operating costs of their internal farm down to a per Ghz-hr. rate.  This enables them to directly compare with our service rates, apples to apples.  In short, unless you only need a few (like less than 10) cheap machines in your farm and run them 24x7, we can always beat the cost of operating an internal render farm.  We feel this is extremely important and relevant to the industry as a whole, since most artists and studios make the (incorrect) assumption that rendering on their own farm is "free".  No one can compete with our power, cooling, hardware, software, and facilities costs, which is what makes our service so cost effective.  Let me know if your studio accountants or owners would like to get a copy of this spreadsheet.

More updates as they come.

[digitalis99]

Hey TG guys, just a quick update on our farm.  We've dropped our Ghz-hr rates by $0.01 across the board.  This means our lowest priority render rate is now just $0.01/Ghz-hr.  That's right, 1 penny per Ghz-hr.  We still have no minimum, no pre-pay requirements, or any other stipulation to prevent you from getting this rate.  It's available to everyone, everyday.

Our focus on efficiency at every level of the render farm process has proven very effective in the past year.  We've dropped our rates to pass on the cost savings to our users and maintain our position as the price/performance leader.

[heriosan]

It's great to hear about all of your updates and upgrades! I would really like to try your service, but unfortunately we are all Mac at our studio. Has there been any development on an osx version of your application, or any work arounds?

Cheers