Also there are snapshots of what she's supposed to look like, when things are working correctly, here: here's one for example; https://github.com/opencog/blender_api/blob/master/docs/eva2b.png I think the current textures are for the one on the right side.
I feel as if I'm bragging and showing off, but exactly 0% of this is my work; it's more like a "see, this is what it looked like 7 years ago" and to me, it looks pretty darned good. --linas On Sat, Mar 5, 2022 at 9:13 PM Linas Vepstas <[email protected]> wrote: > Hi Reslav, > > On Thu, Mar 3, 2022 at 8:26 AM Reslav Hollos <[email protected]> > wrote: > >> Hi Mark, >> >> I've added "Principled BSDF" node materials and approximated the values >> to match the look, >> > > It should not require any "approximation", the skin texture is there, as > well as several different eyeball textures. > > There's the scary-looking > https://github.com/opencog/blender_api/blob/master/textures/eva.COL.png > and then a "subdermal" evaSubdermal.COL.png and then some skin gloss > texture eva.GLOSS.png and perhaps a skin-transparency (so the subdermal > shows through??? guessing wildly) in eva.DISP.png .. there are textures for > the inside of her mouth, the eye.png but also eyeCornea.png and finally, > long long eye lashes.png > > I spent multiple sessions earlier this week, trying to figure out how to > wire it all up and was completely stumped. > > I can see that original blender file access these textures. I can also > see many dozens of animations, but again, was stumped as to how to make > them run. I think its important to port these, because they all looked > pretty good (to my untrained eye. Certain cough cough sculptors in the > house were less than happy.) The only "major" flaw was that if you used > the various viseme animations (visual phonemes) while she's talking, and > then made her smile, the mouth would go weird. > > --linas > > >> except the eye's displacement (if you zoom in really close) which I've >> not figured out yet. >> Shadows/lighting are not tested, I can continue on this in a few days, >> but hopefully this would be sufficient for you to start getting some >> visuals. >> >> The blend file is here >> https://github.com/Radivarig/blender_api/blob/update-materials/Eva_nodes.blend >> >> Best, >> Reslav >> >> >> On Wed, Mar 2, 2022 at 5:23 PM Mark Wigzell <[email protected]> >> wrote: >> >>> Hi Reslav, >>> Hey great! Thanks for your help: >>> In order to see issue: >>> >>> git clone https://github.com/opencog/blender_api.git >>> cd blender_api >>> blender -y Sophia.blend -P autostart.py >>> >>> You can download different blender versions from >>> https://download.blender.org/release/ >>> If you use 2.79b it seems to work best. If you use the next version, >>> 2.80rc3 you'll see a lot of load errors. >>> My actual work on Eva (Sophia.blend), is around getting the existing >>> solution to build and run. You can see that here: >>> https://github.com/opencog/docker/tree/master/noetic >>> Try building ros-base, ros-blender and ros-eva containers. Finally run >>> the ros-eva container, it comes up using "tmux". There are run errors in >>> some of the nodes, the webcam image comes up, and the current blender (3.?) >>> comes up with Sophia.blend. Of course it looks terrible because the >>> materials haven't been converted. >>> I attached an image of the 2.79b running. (If you click the "start >>> animation" button on the right, she goes into action.) >>> Apparently Eva should not have been renamed Sophia.blend, that was a >>> mistake 7 years ago or whenever. I'll fix that. >>> So running Eva in blender with the "blender_api" in the current working >>> directory is essential to having the ROS node work properly. And having >>> that work properly is essential to debugging the "perception" repo, which >>> won't work without blender ROS >>> >>> - There is an issue of python namespace when starting up Eva in >>> blender. >>> - There is also one "cyclic dependency": Dependency cycle detected: >>> deform depends on control through Child Of. control depends on deform >>> through Locked Track. >>> >>> On Wednesday, March 2, 2022 at 12:01:22 AM UTC-8 [email protected] >>> wrote: >>> >>>> Hi Mark, >>>> >>>> Is your current progress available somewhere? I can look into Blender >>>> materials. >>>> >>>> Best, >>>> Reslav >>>> >>>> >>>> On Tue, Mar 1, 2022, 18:27 Mark Wigzell <[email protected]> wrote: >>>> >>>>> Hey guys, well I have spent a couple of weeks re-building >>>>> opencog/docker to Ubuntu 20.04 and ROS "noetic" release, concentrating on >>>>> re-vamping the "eva-ros" container initially. >>>>> I ran into various issues with old python code, easily upgraded. Also >>>>> some issues with conflicting configuration parameters, I think those are >>>>> all solved or solvable. >>>>> >>>>> The issue that has become insurmountable is the change in the Blender >>>>> implementation from 2.79b to 2.80rc3. The key change was they >>>>> dropped support for the Internal Renderer in favor of Cycles and later >>>>> Eevee. >>>>> >>>>> I have tried to convert the textures using the Materials add-on they >>>>> supply, but it failed to convert the eyes, and was less than perfect on >>>>> the >>>>> skin tones as well. If this one issue could be handled by a skilled >>>>> Blender >>>>> user, we might be able to "save & load" up to the current Blender release. >>>>> >>>>> The current Blender release has the distinct advantage that it runs in >>>>> an Ubuntu 20.04 docker container "out of the box". The older blenders, >>>>> including 2.79b do not. There has been a lot of churn with X11 >>>>> acceleration, 32bit to 64bit, and the new Wayland support, and the default >>>>> Ubuntu 64bit setup confounds the older blender versions. >>>>> >>>>> With a properly running Blender, the Eva project could be re-vamped. >>>>> Is it worth it? All I can convey is my enthusiasm for the project when I >>>>> trailed it back from the "Sophia the robot" interviews. Whilst I >>>>> understand >>>>> that basic research into AI learning does not need the heavy load of >>>>> supporting robots or human like interaction, I strongly believe that >>>>> having >>>>> such a teaching example fully functional would do wonders for the entire >>>>> field: many people are inspired by the hype around AGI and androids, >>>>> >>>>> Providing fledgelings with a setup such as was developed for Eva would >>>>> be a powerful kick-off for anyone wanting to dig into the field, >>>>> especially >>>>> young people, who would definitely get a kick out of running their >>>>> prototype AGI hookups in the Eva rig. There IS something inspirational >>>>> when >>>>> she comes to life, breathes, smiles etc. Granted the rig is NOT itself a >>>>> demonstration of AGI, but the hookups back to the AtomSpace should allow >>>>> people to make all sorts of experiments that point in that direction. >>>>> >>>>> I think the Eva project is a great landing place for enthused talent. >>>>> It just needs to get over this current hump, which has defeated me at the >>>>> moment. >>>>> >>>>> Perhaps someone in this group has access to some Blender people we can >>>>> engage to remedy this situation? >>>>> >>>>> Regards, >>>>> --Mark >>>>> >>>>> -- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "opencog" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to [email protected]. >>>>> To view this discussion on the web visit >>>>> https://groups.google.com/d/msgid/opencog/CA%2Ba9A7CcAcywy0MqqzjG7uf-zi%2B6OsPvt%2BWiWR2YbH3poarJxg%40mail.gmail.com >>>>> <https://groups.google.com/d/msgid/opencog/CA%2Ba9A7CcAcywy0MqqzjG7uf-zi%2B6OsPvt%2BWiWR2YbH3poarJxg%40mail.gmail.com?utm_medium=email&utm_source=footer> >>>>> . >>>>> >>>> -- >>> You received this message because you are subscribed to the Google >>> Groups "opencog" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/opencog/ccc3368d-10bf-4313-9f3f-d724c5632b44n%40googlegroups.com >>> <https://groups.google.com/d/msgid/opencog/ccc3368d-10bf-4313-9f3f-d724c5632b44n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> >> >> -- >> Kind regards, >> >> Reslav Hollos >> >> -- >> You received this message because you are subscribed to the Google Groups >> "opencog" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/opencog/CANHDs8ncAzQgtkJE%3D633XKUNstPAv41A7r%2B-AXVuQ13OEVeNZA%40mail.gmail.com >> <https://groups.google.com/d/msgid/opencog/CANHDs8ncAzQgtkJE%3D633XKUNstPAv41A7r%2B-AXVuQ13OEVeNZA%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> > > > -- > Patrick: Are they laughing at us? > Sponge Bob: No, Patrick, they are laughing next to us. > > > -- Patrick: Are they laughing at us? Sponge Bob: No, Patrick, they are laughing next to us. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAHrUA36Ln8NUCOtjSvNJtDfH2L%3DJgnTs6CpdL3o9%3D42yNn%3DJXg%40mail.gmail.com.
