Jump to content

davpe

Members
  • Content count

    385
  • Donations

    0.00 CAD 
  • Joined

  • Last visited

  • Days Won

    4

Everything posted by davpe

  1. FOR EACH LOOP / SPLIT PRIMITIVES?

    you may want to watch this video:
  2. Render is upside down

    heh, this upside down thing is the funniest bug, everybody who see it for the first time goes totally nuts about trying to understand what the hell's going on
  3. for reflections you want to look at Reflection masks (on receiving object). for shadow you can remove shadow render visibility (on casting object). both can be found in object render properties. for primary visibility you can mark objects as phantom.
  4. thing with arnold is that up to arnold 5 there was no adaptive sampling so you could predict render time as you described. disadvantage is that you're shooting specified amount of rays per pixel, regardless if it's needed or not. with adaptive sampling on you won't be able to predict as the final number of rays fired isn't determined beforehand (there is just a certain threshold). in Mantra, you can turn Ray Variance Antialiasing off and you get this simplified thing where you can say 1 sample - 5 mins, 2 samples - 10 mins. in some cases you can benefit from this brute force approach but typically you won't.
  5. that's a good question. I think you should benefit every time when bump features are smaller than a pixel. can't say for sure thou.
  6. that would interest me too. it is not anything new, something similar has been implemented in V-ray a few years back (not sure how good it is as I'm not a Vray user) and I was a bit surprised then that nobody seemed to care much. no clue how would I put it into practice but I really hope to see this built into default Mantra shaders in the future.
  7. we've just finished a project involving several digital doubles (closeup shots, some curly hair, long hair and braid). mainly for the reason of having not enough houdini artists available for the project at that time, we decided to use maya and xgen. at some point, we ended up exporting some of the haircuts from maya to houdini, finishing the groom there and exporting back to maya for rendering (just because everything else was already done in maya). pointless to say this was far from a smooth workflow but with xgen, some things we were unable to do. for the next project, I think we'd rather go entirely houdini way. my personal opinions: you might be fine with xgen if you have maya artists that really know how to use it and are aware of all it's pitfalls. it is definitely easier and faster to start with and deliver first look iterations. however, there are some unfortunate limitations and bugs (i guess) that can make more complicated haircuts and especially changes late in the process very frustrating and time consuming. also, you have to count with the possibility of loading a maya scene and finding out that your xgen hair is gone (and your own hair too after seeing that ). very frustrating. with houdini, you'll probably need some more time to figure things out - hair workflows there aren't as straightforward as in maya and you might miss some convenience tools at current time. it may be also a bit slower in terms of viewport interactions than xgen. generally, it's like everything in houdini. more difficult to master and you'll need to spend more on hardware, but you'll never run into situation where you aren't able to push something further unless you make the whole thing again. complicated haircuts (long hair, braids), interactions and changes (!!!) can be handled much more easily. transferring a haircut to a different head also seems to be quite easy (although we haven't done that, so don't take this for granted), so it is potentially huge time saver if you have more characters with similar hair or if you're doing hair stuff frequently. if you're thinking about building solid hair pipeline i'd opt for houdini as it is much more flexible and seems to be more future proof than maya. cheers.
  8. the more you know the better for sure but if you're mostly hobbyist, you don't have to care much. just do what you like houdini is my no.1 choice too as it is most open, most innovative and seems to be the most future proof general purpose 3d software today.
  9. it may feel overwhelming looking at a ton of nodes in your scene and more complex work definitely needs some more planning than what's usual. great advantage thou is that you can go back anytime to any stage of your modeling and change things, and those changes will propagate through the network. with some practice it is a very powerful thing, offering a great deal of flexibility. although at first, this behavior can be quite infuriating... but hey, that's the same feeling like first weeks in zbrush!
  10. i use MSS all the time. it is extremely handy for material overrides, for example (when you have more mantra nodes and each of those apply a different style sheet), or if you want to render packed primitives, MSS makes it easy to assign materials without having to unpack the geometry (so very heavy scenes can be kept relatively lightweight). also, there are several videos online that show how to do quite complex material setups for crowds. in general for scenarios where you have a lots of objects to which you wanna apply the same general look but with variations, MSS are great. all of these things can be worked around to get the result but MSS sort of unifies the workflow and I'd say it also makes it easier to do. another thing is that it allows easy sharing between scenes as it is a json script that can be exported into text file and also edited as a text file if you wish - very cool for big projects. plus it offers more functionality i haven't even discovered yet.
  11. yeah, zbrush approach to modeling feels actually completely opposite to houdini. and I love both for it's uniqueness btw. houdini's vdb modeling tools is actually sort of what zbrush is doing. also, you might be interested in looking at this: https://gumroad.com/l/GVLLS
  12. hi, 1) this is what Create in object level / Create in context is doing (dropdown menu in the top right corner of your viewport window, right next to the question mark icon) 2) bummer. don't know, I rarely use shelf tools or viewport selecting. it's odd. 3) not sure what the end result should be but this is principally not working. first of all, your groups selection are not intersecting so it is doing nothing, as expected. also, your meshes are quite messy with some bad topology issues. not good for booleans. if I converted to vdb meshes then it works. check the attachment. PC_response.hip
  13. Mantra - Roundededge Node

    True displacement off means just bump. I suspect thou there is still displacement calculation going on if it is plugged into displacement (not sure by that, you have to test it). A: that's right B: yes current geo doesn't have enough resolution. if you can't subdivide to get enough resolution you're left with option A if you want to stay in houdini C: other solution would be to use texture painting software
  14. Mantra - Roundededge Node

    ok the thing is that displacement is evaluated first and then edge detection is performed (as it takes a new geometry created by displacement into account). easiest solution would be to A: bake the mask into a texture first, or B: do it in SOP level and bring it into the shader as an attribute.
  15. Mantra - Roundededge Node

    i can't really see what's going on there... something is definitely screwed up in your setup. can you post the file?
  16. I haven't ever used c4d, however in last 2 releases or so, houdini made a big leap forward in modeling workflows. it is true that there are still some functionality missing but its mostly convenience stuff, everything essential is there. Personally I love it for how flexible it is and how it never gets overwhelming with very complex models. very complex means just lot of simple things, not one massive hard to manage asset. That also makes changes to the model relatively painless and much less time consuming than in other packages (if you laid off your stuff well). It requires more technical approach thou so it depends what you're looking for.
  17. Mantra - Roundededge Node

    i use it quite frequently and it works fine for me. not sure what are you trying to accomplish in your setup, however what i can see at the image doesn't make much sense to me...
  18. I'm running 16.5.405 and it's not happening here. i'd send a report to sidefx. looks like a bug.
  19. correct me if i'm wrong but i think alembic doesn't support anything else than polygonal geometry with standard attributes (like N, uv...). no instancing, no packed geo, no custom attributes, no anything. And I guess it doesn't really matter as other software packages wouldn't know how to read it anyways.
  20. Rendering time

    looking at the pics I can only tell u there must be something terribly wrong in your scene or render settings as something like that should render in a minute or so. what are your render settings and what materials are u using? is it still slow when you render just a plain grey shader? what kind of lights do you have in your scene?
  21. Rendering time

    you can render low quality, just to demonstrate what you've got
  22. PBR Translucency

    I suggest you to use Classic shader as that has translucency shading implemented (as one of refraction modes). Also, you may want to use single scatter SSS rather than translucency (renders faster, looks better sometimes).
  23. Rendering time

    hi, this is such a broad topic that it cannot be answered just like this. post a scene, post some renders, ask specific questions.
  24. Allocation memory problem

    well, it seems like you've got insufficient memory. a good way to solve the problem is to add more memory cheers
  25. copy to points

    what you can do is to create a group of template points where you want to have doors (by bounding box for instance). then based on group membership, override the attribute number you've created before for windows (so that all the door template points have the same number). then if the attribute number is the one you've generated for the doors, pick a door shape instead of window. that's in short how i'd go about it.
×