Announcement

Collapse
No announcement yet.

Deadline 10 & Vray Render History / Autosave

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Deadline 10 & Vray Render History / Autosave

    Hi,

    We have been using Deadline with 3dsmax & vray more frequently over the last few weeks for stills in order to utilise all available resources within the office - one thing that seems to be stumping us is that the VFB History Autosave is completely disabled when rendering via deadline. For obvious reasons I can understand why this would be disabled for animation sequences, but for still images having the option to enable the history would surely have advantages for certain workflows.

    For example: As I'm sure is the same with a good number of smaller practices, a high percentage of our workflow is in the form of stills & our current workflow seems to be quite monotonous...

    1 - Set up first camera & finish test rendering - including adding all lens effects & corrections
    2 - Remember to head over to Render Setup & add in the location to save the vrimg
    3 - Deadline config & submit (head back to step 2 as we almost always forget under tight deadlines)
    4 - Change to next camera & finish test rendering - including adding all lens effects & corrections (remembering to disable the above vrimg save location)
    5 - Remember to head over to Render Setup & re-enable the save the vrimg (& also remembering to update the file name accordingly so that we don't get conflicts / overwrites)
    6 - Deadline config & submit (head back to step 5 as we almost always forget under tight deadlines)
    7 - Repeat the above for remaining cameras in our scene
    8 - One way or another get the vrimg back into our vfb history to ensure they are ready for region rendering once we have passed client comment stage (This is the one that causes the most issue as the corrections are not saved so they either need copying and pasting over from a previous test render if we kept one, or alternatively re-implement them from scratch. Also... the vrimg's saved from deadline drop to the bottom of the history as they are not sorted in chronological order, they have no file names or additional detail such as camera render time etc which is a real pain in the backside for later house keeping)

    Obviously, if the VFB History Autosave was functional, it would be as simple as set up your camera, submit via deadline & then see the images re-appear in your history once complete - the only consideration I can imagine is that Deadline will need it's own overriding config options for the Autosave location to ensure that there aren't any conflicts between nodes (I expect that these settings would be sticky within the deadline submitter as they wouldn't need updating from job to job - And would also be accompanied by an enable / disable button which would be completely disabled if the job has an image sequence assigned).

    I'm raising this on the Thinkbox forum as well as on here, mainly as I'm unsure where the current restriction comes from. Fingers crossed someone can either point out the error of our ways or give us a quick fire solution to enable the render history.

  • #2
    +1 for this

    This is the reason we hardly ever use deadline or any other render manager other than VRay's built in distributed rendering. The ability the quickly save, load, and adjust vrimg from the VFB history on the fly is too valuable to us as we go thru many revisions.

    Comment


    • #3
      Originally posted by AbeTJH View Post
      +1 for this

      This is the reason we hardly ever use deadline or any other render manager other than VRay's built in distributed rendering. The ability the quickly save, load, and adjust vrimg from the VFB history on the fly is too valuable to us as we go thru many revisions.
      Yeah, so we're not alone - I forgot to mention that we've been exploring other options in an attempt to move away from distributed rendering. Deadlines power management (via pulse), admin console & comprehensive failure detection makes it an extremely attractive option if we can just overcome this small hurdle!

      Comment


      • #4
        I agree, to have a VFB History is a must in many cases here at Brick as well. Therefore we implemented the feature in Pulze Render Manager to be able to send out a distributed render to the farm and having the master node the workstation in front of them. In that way the VFB History can be saved automatically. On top of that in the final release in February we are planning to extend this feature and allow the user to save the vrimg file from any kind of render jobs.

        If you are interested download the open beta and give a try: https://pulze.io/products/render-manager

        Comment


        • #5
          Originally posted by Attila Cselovszki View Post
          I agree, to have a VFB History is a must in many cases here at Brick as well. Therefore we implemented the feature in Pulze Render Manager to be able to send out a distributed render to the farm and having the master node the workstation in front of them. In that way the VFB History can be saved automatically. On top of that in the final release in February we are planning to extend this feature and allow the user to save the vrimg file from any kind of render jobs.

          If you are interested download the open beta and give a try: https://pulze.io/products/render-manager
          That sounds somewhat promising, however it appears that jobs currently have to be submitted from directly from the render manager - do you have any plans to incorporate a submission ui directly inside max? Multiple still cameras within one scene file would obviously cause some work / potential issues.

          Cheers

          Comment


          • #6
            I have a good news For multiple camera setups we have a separate tool called Scene Manager. Render Manager automatically reads out all your setups and output info from the Max file and you can batch render them easily. You can check it over here: https://pulze.io/products/scene-manager
            Yes, jobs has to send out from the standalone software. In that way you can create jobs from browsed files without opening them at all. This stand alone structure allows us to keep the system flexible and capable to handle different tasks and DCC-s simultaneously in the future. May I ask your opinion on this? What is your argument on sending out jobs directly from Max?

            Comment


            • #7
              Yeah, we spotted your scene manager & it definitely looks great in it's own regard, although again it doesn't really fit the bill for us (great for local rendering / dr etc but essentially we're after a render farm manager with flexible power management & failure redundancy that I don't believe this can offer).

              Our 2 cents on external submissions: Call us lazy, but saving out scene files and heading to other software to render just sounds like too much work when we're already juggling pushed deadlines and multiple iterations - having a submitter from within the host UI also gives us the option to choose cameras and tweak settings on the fly.

              Comment


              • #8
                Yeah, we spotted your scene manager & it definitely looks great in it's own regard, although again it doesn't really fit the bill for us (great for local rendering / dr etc but essentially we're after a render farm manager with flexible power management & failure redundancy that I don't believe this can offer).
                Thank you for your kind words about SM. The Render Manager is unique in a way that it automatize the utilization of the farm. It is automatically checks the existing jobs and always distributes the available capacity between them. In this way the utilization of your farm is maximized and you save a lot.
                Of course for that we have a powerful software, hardware, plugin and sanity check under the hood. It is a must to be able to guarantee the streamlined operation of the automatized system.
                Please do not misunderstand me it is absolutely fine that you believe in Deadline, it is a great tool! I just wanted to let you know what was our experience in the last couple of years at Brick Visual and where we ended up.

                Our 2 cents on external submissions: Call us lazy, but saving out scene files and heading to other software to render just sounds like too much work when we're already juggling pushed deadlines and multiple iterations - having a submitter from within the host UI also gives us the option to choose cameras and tweak settings on the fly.
                Just a minor correction: You do not have to save out anything. The Render Manager automatically recognizes all the running Max instances and scans them without saving the file. So basically the workflow is like drag a running window on the top, send out the job and continue the work. It takes the same time and effort like opening the render settings in Max.

                Comment


                • #9
                  Originally posted by Attila Cselovszki View Post
                  Just a minor correction: You do not have to save out anything. The Render Manager automatically recognizes all the running Max instances and scans them without saving the file. So basically the workflow is like drag a running window on the top, send out the job and continue the work. It takes the same time and effort like opening the render settings in Max.
                  We didn't pick that up from your website - game changer! As soon as we've got some time to test, I'll sign up for the Beta & check it out. Thanks again!

                  Comment


                  • #10
                    Originally posted by Attila Cselovszki View Post
                    I agree, to have a VFB History is a must in many cases here at Brick as well. Therefore we implemented the feature in Pulze Render Manager to be able to send out a distributed render to the farm and having the master node the workstation in front of them. In that way the VFB History can be saved automatically. On top of that in the final release in February we are planning to extend this feature and allow the user to save the vrimg file from any kind of render jobs.

                    If you are interested download the open beta and give a try: https://pulze.io/products/render-manager
                    Ok. I finally managed to find time to download the render manager and after a few teething issues I managed to get it all up and running. It has left me a little confused though: the frame buffer history is completely disabled just as it is with other render farm managers - I was under the impression from your fist post that that wasn't the case (I did spot that we can save a vrimg independently just as we can with other software but that doesn't solve the problem in hand). Can you let me know if I'm completely missing the trick here?

                    Comment


                    • #11
                      Ok,

                      So we've explored this as far as we can with the Thinkbox team, basically what we know so far:
                      1. If we start the render from the smtd with "force workstation mode" enabled, then the history works as expected - flip side of this is that each workstation has to have the same vrayvfb.ini settings to ensure the vrimgs are all saved in the same location & further to this, this approach simply won't work with any node licenses.
                      2. If we render without the "force workstation mode" enabled (effectively in slave mode) then the history doesn't show, nor is it activated / configured
                      3. If we pass "vfbControl #showhistory true" via deadline, then the history window show (albeit the settings are still disabled) - clearly shows that we have some form of communication via deadline.
                      4. Obviously for whatever reason vray doesn't call the vrayvfb.ini settings when initiated in slave mode, nor do we have the necessary settings at our disposal to configure it manually (at least from what we can see on https://docs.chaosgroup.com/display/...ogrammatically)
                      Can anyone chime in here(possibly even one of the Chaos team) to let us know how to enable, or configure the frame buffer history when rendering is slave mode - it would be a godsend if we could get this up and running.

                      Cheers

                      Comment


                      • #12
                        Hi, Sorry for the late answer I missed your question about Pulze Render Manager. So I just informed you about a possible workaround when you send out a DR using your PC as a Master node. In this case the VFB history can be used without hesitation.
                        As far as I understand you are looking for a different workflow. No worries, just please write us to support@pulze.io and let's discuss the details. Your request seems feasible and we will be able to implement it. We tested it and it looks promising.
                        Cheers

                        Comment


                        • #13
                          Hey, no worries. I'll get an email across to you later today

                          Comment

                          Working...
                          X