Maketx make my life simpler?

I have a scene with 12gb of video, comprised 6500 frames as ifl sequence.
I am getting crashes after around 600 frames - and that is only test res, so I’m concerned.
Yes I can split them into chunks but would rather not, with so many to do.

I thought of converting them to tx but that procedure alone will take a very long time.
It also makes a 600kb file 11mb, which is why I ask; is this a worthwhile thing to do and will it solve my issues with memory
usage in this case?

Thanks for any help with suggestions.

Why the scene crashes can’t quite be determined from what you wrote.
However, the .tx will *always* make a scene render optimally, memory-wise.
Size of files on disk, particualry when comparing a jpg with a compacted, atlased texture, are meaningless.

Ultimately I reasoned that the time to even convert these files (around 260,000) would be silly, so I chose another method.
Also I discovered that only when all the textures were visible in the camera view was it really a problem, so I could only do around 30 frames per batch, of a e.g. 200 frame seq.
As the shots were not all like that, I managed to use backburner to render all 70 shots relatively quickly and without issue…just a bit of babysitting.

The interesting thing about backburner (and that which sealed the deal) is that I could simply stop and restart it, (which takes seconds) when it reached peak
and it would clear the ram. Without BB I would have to reboot Max each time.

So my remaining question, out of interest, is why Max doesn’t clear the ram when I stop the render normally, without using BB?

Because it’s Max. :smile:
Deadline offered the option to restart the “renderer” (i.e. the DCC carrying it) since day one. :slight_smile:

Important to not that if you used .tx inside vrayBitmaps, you’d then be able to tell v-ray to unload the textures between frames, directly.

Lastly, is that 260k texture file that are going to be thrown away after this project?
If that’s the case, you should perhaps invest in a dedicated box doing just .tx conversion for you, and you’d still be saving a metric ton of money over the course of one such project.
However, if your scene didn’t actually carry 260k textures, you can convert just the ones the scene uses quickly, with the provided right-mouse menu tool.

LOL yes that is my thinking…they missed something there I think :stuck_out_tongue:
I did look at using Deadline but tbh even the process of installing it was a complete nightmare, only to find out that it is now only a paid option and the prices are ridiculous,
needing various additional paid parts for it to function. I wasted a good 3 hours installing it only to discover this, as the process lacks transparency re. pricing.
Will not go near it again.

Wow, really?
It used to be quite straightforward, and definitely free for personal use (no idea on prices for the commercial though.).
I feel sorry for the devs if that’s the case, they are exceptional in many ways, they don’t deserve the extreme mercification of their baby.

EDIT: on the site it states it’s free for up to ten nodes. Have they made a mess of the install/registration/licensing through amazonification?

Yes it was the process of having to go through Amazon web bollocks. It was confusing and ultimately led me to a payment portal which
listed the various things I needed to have, including not only having to pay for the render time itself, or so it appeared, but also having to pay for the link to Vray and possibly other things.
I just gave up…seemed contadictory to what I had heard, which is what you suggest, that it is free to a certain degree.

I think at some point you may have ended up to the amazon cloud, eheh.
I’ll perhaps poke the guys and let them know the process isn’t ideal, for the little it’ll help.

Yep, that’s where I ended up.
The only other link info on pricing is this AWS Thinkbox
which leads to a form, leading to waiting for someone to look at the form, contact me…then after that I again lose interest as it is non-transparent
and seemingly a way they think to hook you into other things…yes, I’m suspicious :stuck_out_tongue:

I think my main gripe is that I wanted to install this, thinking it was free, to just discover if it would indeed do what I wanted (what ultimately BB did simply).
Then I would have gladly paid the $48 for a year’s use, just to get this project out of the way.

However, it took so long to eventually get it installed and even then that installation/setup process was bewildering, ultimately pointless and seemed overly complex for what I imagined the
thing did. I’m probably not their intended user and that this is more tailored for use by a sprawling large company who have admins to manage this boring nonsense :stuck_out_tongue:

Quite strange you had a bad experience with Deadline. How big is your renderfarm? They recently raised their free license nodes amount to 10 nodes. Also, I recently installed deadline to render some things and it was quite fast and straightforward.

I only have the one machine…So it’s more like a petting zoo than a farm :stuck_out_tongue:

Maybe I just got led down the wrong route for its installation…it seemed there was only one, via installing all the Amazon portal stuff.

Incidentally I think I did discover that it is not useable with only one machine as slave and master, so anyway pointless at this point.

Please link to the correct method if you can, for future reference :slight_smile:

go to AWS Thinkbox , click “download Deadline”, log in to Amazon in the next page.
At that point, you will land on https://downloads.thinkboxsoftware.com/#/, and all you have to do is click to download deadline 10.1.17.4, a 1.2Gb Zip file.
Run the Repository install (You’ll need also to install the MongoDB. Repo and Db will store and deliver the job data -both DB entries and physical scenes- to clients).
As you will run the whole thing on the single machine, feel free to skip the authorisation part.
At this stage, you installed the Deadline “server” components.
Run the client Install, installing just the Client, as you won’t remote connect to the server, but use the file system.
Now you can install the submitter plugins and scripts from your repository folder (.\submission\3dsmax\Installers).

Open max, notice the deadline menu entry, submit away.

If all is good, the worker (AKA slave) will pick up the job.
You can check on this by launching the Monitor from the submitter menu.

I probably missed some specific bit here and there, but this is the gist of it.
It does run on a single machine, and it does not require the AWS cloud rendering part to work, that i could see.

Thanks Lele.

I definitely installed it, albeit via the annoying other route with the whole Amazon stuff. I remember it telling me that it wouldn’t run on a single machine as slave/master so maybe I installed a completely different version (the cloud one)?
No idea. Very odd indeed.

I’ll give it another go when I get a moment, using your link (as I uninstalled everything anyway), purely to see if it can do the same as what BB ultimately did very well. :slight_smile:

Well, now I feel stupid. I am always using the scene converter for older scenes. But I never paid attention to the menu item below it. This is great.
I hope there will be a nice little tool like this to converting and renaming textures for aces in the future.

I am technically illerate, really I am and I set it up in a few hours, following only the Amazon web instructions and one youtube vid.
Free for 10 nodes regardless of CPU / GPU count.

Def worth the time. I cannot believe commercial businesses still use Backburner. I know a handful of backwards arch vis businesses that persist with it…tragic.

Well yes, I didn’t like BB over the years, as there were always issues with the setup.
But for this particular project I just needed something that would solve the specific issue I had, which BB did, plus for whatever reason, I had no issues with its setup.
The last time I used it in a studio setting with many machines was a long time ago and Deadline or similar did not exist (or was under my radar/cost-prohibitive) at that time, so no choice really.

My general motto is ‘whatever gets the job done’, so, whether generally crap or not, BB totally solved my problem on this occasion :slight_smile:

So the first thing I noticed is that the version I had downloaded/installed was 10.1.15 and with the installer you linked @lele it installed in a far simpler way (thanks also for your ‘idiot’s guide’ instructions which certainly helped this morning…a tad hungover :p) and I now have it working. So YAY!

Thanks also to @Morbid_Angel for the mention of the ‘restart renderer between frames’ option, which should do exactly what I did want at the time; just about to test this on a problematic sequence.
Although the job is done now, this all will be of use for future work I’m sure.

I do like a happy ending :slight_smile:

Super glad to hear you got it somewhat sorted.
But i’m @_Lele , eheh :smile:

Problems unfortunately.

On submitting the job it runs through a few things in the process but the hangs completely at the stage in the attached.
The only way out is to end task.
I tried on a very simple scene and it works fine, so clearly it has something to do with the vast amount of texture files in the actual scene, as even earlier, after first installing, when I accidentally clicked the ‘Assets’ tab it again just hung whilst presumbly trying to populate that list with all 260k bitmaps rather than just the 45 .ifls that it uses and that the log refers to.
I have tried unchecking the ‘pre-cache asset files on AWS’, plus choosing to ‘save and use current…network path’ but the same hang occurs.

Is there a way around this apparent issue? BB does not exhibit this problem and starts immediately, so this issue has already got me scratching my head, as it’s clearly not only me who will use a large amount of data in scenes.

EDIT: I meant to add that although I can eventually cancel submission, Max still hangs, needing task kill.