ZDG random Daz Studio discoveries and questions.



  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited January 3

    Random top of page distraction.

    left to right, FWSA Wanda costume Chiss eyes Liam hat Sweet Summer shirt costume meme texture, Blooming Pants with AElflaed's textures, Librarian's hair. next is FWSA Krisandra (alien morph) with Zdg Leku WIP (a few pages back), Sweet summer shirt again with a custom meme, Eastern Elegance pants with AElflaed's textures again.  last is FWSA Krisandra (costume Mirialan skin maps) with Kylan hair, meme summer shirt again, and sweet summer skirt with costume colors. Also, some credit to EliteMasterLordofAll (YT channel) and Auralnauts (YT channel) for the cool tunes. Techno Union Anthem (YT vid) by RaKun and Michael Ortega.

    eh, yeah. I had seen that site Prixat. The mesh is ok, it's also nice that it is UV mapped. I wish I could say more nice about it.

    Oh, well, they did offer a download link that wasn't just a STL or max/3ds file, unlike so many others.

    And once the download filter is added, there are only 9 models left,

    and none of them are intact unlike so many that I've seen photos of.

    it was a tad irritating when I was looking last year for a small fun render (or was it 2 years ago, lol. Time can be funny like that I guess, Isaac Arthur day vid). there were thousands of photos of hundreds of mint condition statues of exactly what I was looking for, yet the only 3D meshes I could find were of the 4 broken ones, lol. Maybe in time, the tech will get good enough that scanning these things won't be so difficult. And yeah, I was looking for both the sitting and standing ones.       Oh, in fairness, that game consol I did get from there, so I had seen the site before and it's not all bad. there is some good stuff there and I do appreciate the efforts of the caretakers at the museums.

    P.S. today's Isaac Arthur vid, may have a good explanation of how coffee is so powerful, lol.

    2564 x 1416 - 375K
    1066 x 1352 - 241K
    1066 x 1352 - 550K
    3200 x 768 - 731K
    1200 x 960 - 107K
    2800 x 2100 - 2M
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited January 16

    k, I had put in for a few ships, tho there are plenty more to pitch in for if ye want to see it on Sam's wall over at Trekyards (YT vid),

    Trekyards 'had' a what ships would you like to see live.The last Livestream was lots of fun with lots of fun chat about different Trek ships, and while I am not able to sponsor more, I do hope to be there in the chat for the fun.   As hinted above in the former render, I do have another set of figures open on the other comp, so I hope to have some fun chat about them. in the meantime...

    In other news, I guess having the 9900k outdo the top end HED part from intel wasn't that great for intel, and having the AMD consumer chip also trounce it in apps the 9900k didn't just didn't go over well. So there are rumors going around about some 10990xe thing (UFDtech YT vid). being rumors, some sound to be the same specs as a former Xeon as if it's somehow a re-refresh meme at this point, others sound like wishful thinking. The angle I'll take, is a wait for more concrete benchmarks and stick with what is on the market now for render boxes being used now.  Not to say, as Steve and Gordon have said, some intel platforms are not doing so great compared to other options out now (GN YT vid).   My 2 farthings, I would not be surprised if that 10990xe thing is a desperate attempt to sell x299 motherboards that apparently are taking up space in warehouses, tho, that is just speculation on my part.

    ok, this may take a tad longer, as I don't have a 320+GigaFLOPS PC (minimum Half a TaraFLOPS recommended) to do spot renders on. This isn't the figure, as years ago I had determined it was something to do with the AoA shader mixed with HD sub-D mesh. I so want to drop a 3950x in that comp.  Overall so far, I do feel 'neutral' about what I've seen so far with the Lekkulian, so far.

    One good side to the AoA over Omni, I don't need to find every single map when moving it to the Daz default. So there is that. And Omni on HD Sub-D mesh isn't much better than AoA on HD Sub-D mesh anyway.

    196 x 110 - 8K
    588 x 110 - 131K
    3840 x 1080 - 1M
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited January 23

    It's been a somewhat sideways weekend. Some things were great, some bad, and some meh.  To start with, the wall of ships is moving along really well and my humble thanks to all that tossed some change into the coffee fund for ships there. I suspect Sam will be rendering ships for the next month.

    Now for the more sobering thing. I had gotten a full-screen notice that updates were over for win7, thankfully Microsoft didn't cause me to lose what I was working on when that happened, the win10 update system still needs some work there. I noticed Joker Reviews had put out a vid on the situation, and before I had gotten moring coffee going, his channel was yanked. A serious 'WHAT!?" moment as I'm sure he didn't do any mistakes like showing a key or something. I later found out that his YT channel had been hijacked and shut down.

    ok, rewriting 3 paragraphs, as this was bad, and it dragged on a bit. I'll let Joker explain what happened (YT vid), and TheQuartering did go over some of the tactics used (YT vid), and SomeOrdinaryGamers also covered it some (YT vid, ah, language advisory with that one), it looks bad. This distracted me to the point I didn't get stuff done I expected to have ready to post last weekend.

    uh, yeah, I need to get the other renders off the other comp, I think the normal maps are a mistake of a sorts on the guy. Some normal maps can be ok in Iray, yet turn out to be way too 'strong' in 3delight, easy to fix as well. Just select the figure's skin zones, and change the normal maps to 'none'.

    2100 x 900 - 557K
    680 x 380 - 140K
    2800 x 2800 - 2M
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited January 24

    well, after some digging in folders, tabs, and settings, I think it is fair to start this off with a bit of "tempering expectations". The most important being that because of the hands and feet, and other fundamentals of how shaping dials work, combining this with other figures may very well end up with inside out features as I have fumbled across a lot in the past with other products (some also limited to 100% or off akin to the Lekkulian body shape)

    As such, mixing at lower levels simply will result in odd-looking hands, and options of what you can mix in will be significantly limited at full strength for the three-finger hands. Also, the head tails are not an option, they are a must with the provided body dial, as the heads are a tad broken without the head-tail geografts.

    So, what of the head tails, well, they look ok at first glance, however your stuck with the 3 skin tone options included as they do not have an alt torso map UVing option to let you use them with other figures maps.

    Well, that is a hefty dose of tempered expectations, so what's left. Lekkulian G8F with body shape at 50% on the left (note the tiny 4th fingers), and 100% dialed in body Lekkulian G8M right. Outfits, sweet summer left, boys in the sun right (both with costume meme maps).

    well, I can say at least the shaping dials don't break eye pose controls. The pose dials for the head tails do work, yet they do take some fiddling to get the tails to hang relaxed. And the skin maps look ok in 3DL (the 3DL mats are greatly appreciated, thanks  ).

    even with the limitations with kitbashing, the product as is, is ok. I do find it a tad disappointing the hands and feet were not a separate dial for the 3 finger morph, as that would have made the body dial better suited for mixing with other figures. And likewise, not having a separate head and body dial is a limitation. the head tails, well, they work, Abita with limitations and conditions. I like the idea, and the maps and shape are nice for use "as is" no modifying figure pair (and I'm super happy to have a guy and gall combo).

    edit: oh, thats what I forgot to look at, tails on vanilla. I'll toss that togeather with morning coffee as I dig up stats on another 6GB '60' card.

    1920 x 1050 - 616K
    3200 x 1050 - 1M
    1920 x 1050 - 911K
    2800 x 2800 - 2M
    1920 x 1050 - 1M
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited January 25

    loose ends as I sip my first cup of coffee for the day. so, Lekkulian with the body dialed out, is ok, so from there it is possible to use other figure shapes, as typed before that isn't mixing, that is simply substituting one body shape for something else. At least the border of the head tails geograft is ok with the body shape dialed out, so this could work like this.

    Also, looking at the shaping morphs for the head tails, there is some on G8F, and even dials for the left and right independent.

    On the G8M one, there combined, so there is some shaping, their just symmetric in nature, unlike the G8F ones.

    I'm not sure if that was to save effort on a possible limited return on investment thing, or if rawn just got too overzealous with the dial combining (like the combined head and body dials rawn 'loves' to do). it's not the first time that I asked if something will have a separate head and body dial only for the PA to say 'no'. Some minor things that me and rawn agree to disagree on, lol.  Lekkulian does work as is, so as said, it's ok and isn't technically "broken".   segway, ah, yeah. kind of like the RTX2060ko isn't quite an RTX2060super or RTX2060not-super.

    I guess there is a slight difference between 'interest' and 'curiosity'. One implies a hint of liking or want (like my interest in the Lekkulian figures), the other is a bit more wanting to know how it affects the surroundings. The best I can describe the 2060ko is 'curiosity' as from an 8GB 1070non-ti for Iray point of view, 6GB is a bit of a downgrade. The particular attributes that Steve discovered (YT vid) is surprising considering how it relates to the 1660 and 2060 cards.  If you have 'nothing', and need 'something' for blender and other 3D rendering stuff, then all options are on the table, there is that going for the 2060ko. however for gaming, it's meh, and if you already have a graphics card, then it gets a bit more meh. I still need to fill in the graphs I started making back in 2018 for including the confusing mess of 'super' cards, and this just reminded me that I should at least look at the state of things (old incomplete chart)

    in short, as I have a 1070non-ti for Iray, most newer things in my budget range is at best a sidegrade or a downgrade with less 3D environment capacity.  and from a 23watt passively-cooled 4GB GK208 HTPC angle, the 2060ko is a fan-noise watt hog for a juke-box/vid-player just like the AMD 5500XT. so it's meh I think. it's just curious why Nvidia would opt to lobotomize 1200usd or 800usd parts for a 300usd product , I don't think 12nm is that bad with yields. hence the curiosity of what is going on.

    after some digging, I found some stats over at TechPowerup (web page).

    not sure about the prices, tho it looks to be close enough for current trends rather than my usual launch MSRP thing. The 2060ko is an interesting combo, almost like the GTX1070 with 2 less gigs and a few less ROPs. Not everything is in the chart, and I did move stuff up to omit the non-Iray cards. I'll plug this into a few spreadsheets, as I don't have anything current yet on the $/gig and $/CUDA-cores lists. I am courius if the 2060ko's TDP is remotly closee to the 1070non-ti or not.  After thinking about it a second, I should have done 6GB as orange, so that 4GB could be yellow and anything under 4GB as red, oh well, lol.

    1920 x 1050 - 1M
    1920 x 1050 - 858K
    1920 x 1050 - 895K
    1920 x 1080 - 3M
    580 x 720 - 23K
    700 x 720 - 29K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited February 19

    Very interesting few days.

    "VMWare Burns AMD" from a HardwareUnboxed vid made me think of looking something up.

    1,500 USD per four thread node. The first company that licenses 3Delight commercially, in early 2005, was Rising Sun Pictures. The licensing scheme was originally based on number of threads or cores. Since 2018, all purchased licenses are unlimited multi-core. (wiki link)

    so, I hope for those lucky enough to enjoy 64 core Threadrippers (TechGage 3990x web), things should be ok. As for the 2060ko, meh. Think I hinted at it earlier and I'm sure GN-Steve mentioned it in the vid. if you are doing Pro stuff that is not Vram quantity limited then the 2060ko may be worth checking out, if you don't have a Titan or 2080ti already.

    I'm still not sure where the 2060Ko should line in terms of performance in Iray, yet Iray is very Vram quantity limited, VDDR, GDDR or whatever you want to call it. I'm in a bit of a snob position, because for me going from 8GB on the 1070non-ti to 6GB would be a drastic downgrade, and even an 8GB to 8GB sidegrade just isn't worth eating ramen noodles for a year, lol. it doesn't matter if the 2060ko can be faster in Iray, if your scene just won't run on it at all, lol. I find the 6GB cards as of now, are just in that middle ground of too hot to big and noisy for many uses and just lacking capability for the others. Kind of like the latest era of crummy ship CGI.

    *Fun Factoid; it looks like it cost between 88 million to 99 million dollars to make the 11 episodes of the Star Trek Picard show. A SpaceX launch at about 57 million dollars each, literally costs less and they get better video quality.

    Trekyards did look at the Romulan "Stealth Enantiornithes" (youtube link), cool-looking ship, if only the CGI wasn't junk. There are a few other screenshots of the ship, however there covered up by force field effects and other bloom things. It kind of reminds me of some major misconceptions about fanservice.

    "fan service" of late : the act of insulting "longterm repeat customers" for knowing more about a franchise than the studio producing it.

    "subverting expectations" : having a low expectation for something while hoping for better, only for the result to be far worse than could be previously fathomed, lol.

    After episode 2 of Star Trek Picard, I am now wondering if Starfleet, that SciFi entity that used to be the shining example of striving for the best Sentient-altruism you could be, is now guilty of potential War Crimes (ICRC rue55, "The denial of humanitarian assistance"). The basic starting of the show, Picard quit Starfleet because the entire fleet was ordered to Not render humanitarian aid to the Romulans, I find just about everything about that just wrong for Star Trek. I figured it was he just retired for a happy ever after as the beginning plot before seeing the show, so much for happy ever afters, lol. The only thing missing was the shot of Picard drinking a vat of grape pulp as it dripped down his chin, lol.

    so, interesting few days to say the least. Oh, new cute elves. Left SublimelyVexed Aerwyn and middle is Jassii Billie. Aerwyn's mats appear to apply the face bump map to the torso zone (kind of easy to fix), ...

    and the body shape dial makes the eyes do the break pose control rotation (esp at the 50% shape position), hmmm. Jassii Billie, really really nice.

    (edit, 19feb2020, many weeks later. I did finally get around to letting 3DS know about the eyes and bump map mistakes, they are looking into it)

    900 x 600 - 613K
    680 x 760 - 24K
    1400 x 1600 - 1M
    2100 x 800 - 2M
    2800 x 2100 - 2M
    1920 x 1050 - 858K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited February 10

    Yesterday something filtered through the Daz Store that caught my eye, ok, a few things however first I will look at this as I don't have the HDDs plugged in with Pluto and Mercury data. That looks like a fantastic backdrop for Ship-Por-rrr-(uh can I type that here), in any case, the Promos look good for them. On to a project that's been lingering for some time, and maybe a new set in the Daz Stor may be able to augment some of the stuff in my CC0 Zdg Gym set that I spent as much time confusing my pore little mind with FIG documents as I did try(ing, why Grammarly?) to make the mesh do what I needed it to do. So, starting out with looking at the poses in the Z Competition Gymnastics set.

    Overall this does look promising. The first thing I noticed was proper padded endcaps on the balance beam (not rendered yet) with what looked to be proper FIG shaped beam profile (cross-section) in the promos.

    The poses are a nice add-in, much appreciated. I will note that the FIG docs were very confusing with the hights given from the top of the mats instead of the top of the podium for the balance beam apparatus. So I can forgive Zeddicuss for what looks to be a height discrepancy with the balance beam poses. I only made that beam 4 times from scratch before figuring things out on the 5th try, lol. Also guessing from the poses, the beam in the Zeddicuss Competition set may be low on the altitude side.

    Altho it looks like some of the poses were just view-field limb dragged to locations causing the toes to contort in odd ways that result in needing to do a lot of tinkering with some of the poses. More than the usual positioning the figure on the apparatus.

    As for what I've looked at so far, The poses again mostly look ok. The uneven bars look a tad off for the height, however that has changed over the years, so I'll give Zeddicuss that much. I don't know for sure yet if the 2016 document I have is corect any more, I haven't looked in a very long time. I don't know for sure yet if the 2016 document I have is correct anymore, I haven't looked in a very long time.

    The surface zoning and overall detailing of the Apparatus do look fantastic. not quite a perfect 10, or whatever they call it now, lol, it is very good looking even outside of Iray, confusing technical dimensions aside (The PDFs are indeed very confusing, genuinely so for some of the stuff).

    Overall what I looked at so far does 'look' incredibly good, and I will Thank Zeddicuss for not tossing in the pile of old wooden soda pop crates as a Vaulting hazard to Athletes, lol. I have a Free CC0 model of the 'tong' over yonder (OBJ Link in this post, top of page 46 this thread, Daz Content ver here at DropBox) to help fill in the set.  Oh, and a monumental thanks Zeddicuss for the morphing mats in his Gym set, including wrinkles. Drastically better than the former attempt a few years back that had me asking if the PA bothered to try using it first, lol.

    I never finished my Uneven bars, for two reasons. One as ye can see in the screencap, it was incredibly complex with multiple parts that would need the mesh remade from scratch due to my novice attempts. The second being a lack of ambition or urgency out of just not wanting to step on PA's toes. I may never finish that thing I started tinkering with eons ago.

    Besides, the one Zeddicuss made looks really good and doesn't look to be that far off from FIG overall size.

    Oh, FYI, textures used on my set are the AElflaeds cloth, and the Leo is Leotard for G3F Auto Flopped onto G8F, background is Top Of The World. And it looks like there may be a 2019 "Apparatus Norms" document, Hmmm, annnnd it apears the PDF link on the page is busticated, oh well. .. maybe this PDF link works.

    2800 x 2100 - 1M
    1920 x 1050 - 841K
    1920 x 1050 - 641K
    1920 x 1050 - 1M
    1920 x 1050 - 1M
    1280 x 600 - 756K
    1590 x 948 - 809K
    1920 x 1050 - 267K
    680 x 1040 - 39K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited February 10

    Ok, after digging up the current 2019 FIG Apparatus Norms, I guess I still managed to get the height off, opps. lol.

    big cube is 125cm tall and wide, long thin one is 500cm long. My beam is a tad too tall.

    I can forgive the lack of padding on the smaller beams legs, despite it being a rule for competition apparatus, I seldom see the leg padded covers used in training gyms. size aside, that looks damn good Zeddicuss.

    1920 x 1050 - 404K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited February 13

    as I sat down with morning coffee, I sort of got slammed with 2 things, and not exactly bad things. Also, some renders I had going as I slept finished, it is so nice when win10 updates don't reboot the PC mid render before I save them (yanked the cat5 cable, ye is permanently grounded, lol). Oh, and Anna Benjamin Mada or someone at the ZenDesk has fixed the Gothic outfit necklace, the Pendant now loads with the necklace load DUF. thanks thanks thanks.

    And the surface presets appear to work for that as well. (bread crumb. I find it easiest to set up mats, by using the surface zone selection tool rather than the "Universal" one. If you try to click hair, for example, the "Universal tool" will select the figure's head rather than the hair, great for posing, meh for setting hair colors.)

    Over all, looks good. I still need to look into them Aerwyn (Renderosity) eyes, guilty of dragging my feet on that one. it's a far more complex fix I think, than just loading up an item and saving a Daz Studio Asset or whatever it is for a wearable item. I dread just thinking about it, maybe, if I can find the DSF/DSZ/DS-whatever file and open it in notepad++, maybe I can find the offending lines and fix it that way, instead of the ERC-freeze tedium, assuming it's not encrypted, eeek. Renderosity isn't exactly clear about if things got updates, zip named "Aerwyn UD 11 26 18...", Guessing that's 26NOV2018, I will assume the zips haven't been updated yet, hmmm. in any case, I'm rambling out loud, lol. My brother sent me a link to a WTX case that caught his eye, an EVGA thing that may fit the T-rex-40 Threadripper boards (EATX, or XL-ATX).

    And in passing, it looks ok. From the pics, I can see what looks like good vents for a few Tesla compute cards, with enough PCI slots in the case to drop a Tesla in the T-rex-40's 8th slot. I don't think GN Steve has looked at the case yet, tho I think it looks promising for a Desk-side-Compute box, assuming the fans or build quality isn't junk (stuff I can't tell by looking at pics). Speaking of Render Boxes, I got distracted when I was looking for a skybox to use with the gym set, when I noticed the Alvin Bemar Pluto&Charon thing in the Smart Tab.

    Top left is the Explorer class (B5) from ShareCG, top middle is the incomplete Bilskirnir that I was working on, and the Vanishing Point Orville right (that one is a Sweet ship and model). Lower left is one of the billion busticated K't'inga models I came across, no clue where it came from. I appreciate the effort 'someone' had gone threw giving me a version of a D7 and K't'inga, however, that K't'inga was just the D7 with texture maps, and it also had some problems with hovering in mid-air decals. Distractions aside, Pluto and Charon, damn that looks good, even for a "Med Resolution" module (as named in the scene tab). I don't expect to be doing animated low altitude flyovers any time soon, so 'med' is good enough.

    *I added a light to the camera location for the shot of Charon, to make it easier to see in that shot. Oh, and the ships are obviously not to scale, it's more interesting than infinitesimal specs that most would never notice.

    I can also understand completely the complete lack of high res detail of most of the two planets, and I am really grateful actual data was used instead of faking it with a texture brush with completely fictional features. I can also understand others wanting the rest filled in, and I wouldn't be against that so long as it at least matches the low-res data we do have for them. As is, yeah, I'm super thrilled to have a 3D model I can look at from any angle. FYI, the full Pluto scene (excluding the ships I added) only eats about 647MB ish, so rendering with it on non-workstations should be easy enough. As for Mercury, well, I'll say in advance of me pulling it through DIM onto the render box, that I don't expect to be able to get up close on the ground and personal with Caloris Planitia like I want to, I still expect that one to be very nice for more distant ship fly-by renders.

    Oh, I haven't forgotten about the Gym set, I just got distracted, and was deciding if I should confuse things by including assets I'm not thrilled about from other sets. There is an ok looking set of rings in another set, the surface zoning makes it useless outside of Iray with the provided Presets, so meh. I think I'll leave that spot empty on the podium for now. I've yet to do any render of, yet from what I've seen, Zeddicuss did a nice job on the Pommel Horse, looks incredible in spot renders.

    *1, I'm calling them Planets as I have always called them that, and it's shorter than the new 'technical' name. A "dwarf double planet", is still a "Planet". That and after seeing the geological features on Pluto and Charon, I think few outsides of astrology would argue with calling them Planets. There are volcanic features, active glacier regions, and massive mountain ranges, in my book, that is way more than just a rock in space. The rest are technicalities that I would argue also kicks Earth out of "The Planet Club", lol.

    *2, Ship scales, yeah, their way way way off due to me just wanting to toss together a scene to show the Planets.  The Explorer ship should be 5 miles long and change (yes, as long as the Babylon 5 station), about 8046 ish meters. The Orville should be about 2000 meters long. The Asgard Bilskirnir should be, uh, one source stated 1400 meters, another states 1600 meters, and to confuse matters ...

    there are actually 2 completely different Bilskirnir ships shown the first season it was shown in, so anyone's guess as to the version I was working on. "This is getting out of hand! Now, there are two of them", lol.  Charon is about 606,000 meters across, Pluto is about 2,370,000 meters across, with about 19,640,000 meters between them. To scale, the ships would be invisible specs.

    1920 x 1050 - 286K
    1920 x 1050 - 812K
    648 x 507 - 277K
    2100 x 2100 - 669K
    2100 x 2100 - 684K
    1920 x 1050 - 564K
    960 x 260 - 254K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited February 23

    I'm so far behind on stuff.  I need to sit down with a drink and a snack and binge-watch Buildzoids vids as I'm a month behind there, the DDR4 memory Rambling thing and the Ryzen3000 on Gigabyte vid (for 3900x and 3950x) looks really interesting.  Juan (blancolirio) is doing ok and the 737 wire thing is kind of interesting (YT vid). Scott Manely has an interesting YT vid about some almost steam-punk like mechanical rover for Venus competition thing. GN looked into the chewing gum on the 2060ko memory, that was a cool set of tests (no chewing gum was baked to death during the making of that YT vid). And the UFDtech crew is mostly in the US with Bret's son now, things look to be going ok for them (that was a fun vid).

    So, I still have mixed feelings about a 6GB card for Iray, I guess the 2060ko is ok for Blender Cycles that isn't Vram limited.  Off to fuss with cool gym sets, and binge-watch some AHOC vids.  P.S. thanks so much for listening to the fans (longterm repeat customers) Sonic crew.

    Congrats GoG for the fatherhood achievement (doing well vid).  

    The above is for guestimating without actual testing of the thing. akin to my horsepower thing. HP = IPC per thread x clock x number of Threads.

    Hmm, who would have thought clock of the ALU&FPU times the number of operations per-clock for all the Math-Units could possibly hint at something, nope, RPM doesn't mean jack when comparing 4 cylinders 6 cylinder or 64 cylinder engines with different cylinder displacements. "you just can't compare different architectures", sure, lol.  Maybe Wendell over at L1tech (3990x un-testable YT vid) hints that just maybe clocks do mater for more cores when talking CPU gigaFLOPS (wiki) or the more dated MIPS, maybe. The Threadripper 3990x Microcenter build vid Wendell alludes to is here, hour-long and 16min vid ugh... ok, benchmarks start at about 11minutes 30seconds, so not that bad.

    March 2, 2017, Zen turned out to not be yet another APU thing leaving Piledriver and Sandy Bridge to linger on as the high-performance CPU, signaling the end of the 10 years of just 200MHz more on 8-thread CPUs.

    I really want to get a 32-thread 3950x for the GN R7, yet I'm not sure even the 128-thread 3990x Threadripper will come close to that 320 gigaFLOPS number (Just to get AoA subsurface scatter on HD mesh to take less than 5 minutes to compute). The other angle of 'optimization' that Wendell was talking about is not in CPU logic IPC improvements, but in how the software reduces wasted cycles doing redundant stuff. AoA subsurface is really bad, and something in the HD mesh code just compounds that to stupid levels of the FPUs spinning their virtual wheels accomplishing nothing fast.

    (ZDG rant detected)

    Maybe in time, the likes of Adobe and the makers of Marvelous Designer and Z-brush will consider the 'FREE' courses offered by NPACI and the University of Tennessee on writing code and optimizing code for "Parallel computers" that was posted way back in 2002, for use on systems that have been the norm on consumer computers for the past 3 years now. The 9900k is an 8 core 16 thread consumer part, and the R9-3950x is a 16 core 32 thread CPU on AMDs consumer platform, there is no excuse for clinging to the old 1980's single thread single core code for apps made today, especially apps that are not hampered by the Graphics driver draw call thread.

    I don't care what you 'think' the limit is for your Photo processing or 3D manipulation code, I guarantee you there are already scientists and supercomputer programmers around that already figured out a solution to your code difficulties with MPP systems (Massive Parallel Processing), and are eager to share how they figured out the solution. Google search is your friend in such situations, Please join the rest of the computing world in 2020 instead of staying back in the stone ages of 1991 with your CRAY-2.

    sorry, I grow tired of waiting on apps that spin their wheels on a single thread while the rest of the computer twiddles it's virtual thumbs idling, cuz the app wasn't written to be multithreaded.

    Maybe someday loading textures won't be limited to a single thread. as for AoA, having everything else from caustics to reflections to illumination angles take less time to compute than just subsurface scatter on HD mesh is just stupid. here is a random thought, ditch that busticated atomic-level "Solar Photon Random Walk" SSS code and replace it with a simple gaussian blur,

    1. acquire surface illumination map from lights.
    2. apply a gaussian blur to that illumination map with the blur radius dictated by the brightness of the SSS map or setting.
    3. Apply the result to the diffuse maps as illumination values.

    it's a thought. A gaussian blur has got to be a billion times less compute-intensive than whatever the hell AoA is doing on HD mesh, or OmniUber for that matter, they're both horrible on HD mesh.

    To somewhat end this on a funny note, Google was pestering me to look at the new TOS all day yesterday. I saved what I thought was the PDF to look at with morning coffee, needless to say, there was no "brain melting out my ear in a desperate attempt to flee from what I'm trying to do to it" while reading legalize this morning.

     these Google TOS changes are quite like the YT demonetization policies.  They will let me know they exist, but not see what the rules are .  maybe the AI 'algorithm' is taking over.  (oh, he has YT vids and want's to play by the rules, quick, break the links so he can't see the guidelines), lol.

    2800 x 2800 - 2M
    600 x 160 - 23K
    680 x 600 - 99K
    340 x 72 - 27K
    920 x 800 - 354K
    1600 x 640 - 470K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited March 3

    Fun week, fun week. spent most of it vegetating to YT vids, and putzing around in GIMP trying and failing to get a skin tone color masked onto diffuse maps.

    left was Peppermint (G2F) with Alfred's Finery outfit and Kung Fu Hair, right is the not quite anything close Gen6 Aiyana migrated to G8F fumbling ERC-freeze/Transfer-utility attempt ump-teen-million. The Rowana Hair did work out rather well with the TJ8 ears, and the Goth outfit is looking rather nice. The converted figure needs a hell of a lot more work, ugh.

    So, other things going on, hmmm. well, intel finally reduced prices on the xeons, not so much the old ones, just the old Haswell and sandy bridge xeons rebranded with a lower MSRP. At least the mythic 28 core unicorn, water chiller not included, is no longer out in the ort cloud of 15 thousand dollars (2018 GN flashback vid). it also is a bit meh compared to what is on the market today, hamstrung by 14nm stagnation and wafer supply limitations aside, along with debatable 10nm viability if you consider the single laptop part floating around somewhere

    I look at this chart, yes, TechGage did some more C4D testing (website link), well. That 16 core low-end-socket consumer part verses the 18 core High-End-Desktop part, and I seriously am starting to doubt if that 28 core unicorn has a chance against even a lowly 24 core Threadripper 3960x. And that is not even considering the 32 core, 64 core, and 128 core EPYC options. it's not just intel that needs to get off rebranding antique Haswell/sandy bridge architectures with minor microcode rehashes.

    Ask has a monopoly, and it's detrimental to anything beyond single die 7700k CPUs. Apparently, Asetek has a new pump out that has an even smaller microfine area, and that's just not cool (GamersNexus YT vid). Ok, my chicken scratches are not the best before coffee, I know there is no 8780xe CPU, yet there is an 8700k, that's how confusing the intel part numbers are these days, lol. In any case, Most people are not into dellidding their multi-thousand dollar CPUs, let alone gutting brand new CLCs for the fun of it. so they won't know where the CPU die is under the CPUs IHS, let alone what orientation the micro fins are in the round CLC block that can be mounted in just about any orientation arbitrarily on the CPU. They will slap on the bracket, mount it to the motherboard, and rotate the pump cover to be upright if they even care to adjust the logo cover thing. On larger chips like the 7980xe line or multi chiplet Ryzens, most of the chip extremities will be beyond the 29.5mm x 21.4mm micro fin area due to the new pump's fin area being smaller than the dies.

    TR will be even worse with the IO die being directly cooled and the CPU core chiplets off in the corners to fend for themselves for whatever scraps of cold plate they can get.   Oh, and it gets better, lol.   the Enermax Threadripper/xeon CLC apparently has some coolant aging issues (GamersNexus YT vid), hmmmm.

    Props to TechPowerup for the nice photo of that cooler, if only there was a TR/EPYC version and it was available during the Zen2 launch.  GN also just put out a Yeston Cute Case YT vid, looks super nice if you like the Cute Pet or Wife GPU cards from them.  Oh, and a sober note, Freeman Dyson's achievements shall not be forgotten, RIP great one (SFIA YT vid).


    I had no clue this was in the works, the above and former months of ripping out my hair with the transfer utility was not related to RiverSoft's new Gen6 to Gen8 thing. I have had ok luck with the Gen7 to Gen8 thing with Paloma, Dewi, Akimisu, and a few others. So maybe there is a chance the crosseyed wrong eye location or jaw poking through the chin can be worked around. the not round pupils, that may require Z-brush to fix for some. I need to run to the store to get some coffee additive, so I've yet to even read the Gen6-to-gen8 thing description yet.

    The skin maps, that's a sticky issue due to not having a mask to separate out the hand palms, I'll need to get the 3D paint things (blacksmith 3d) and just make the maps from scratch. coffee first, then I'll just plant my head on the desk and sob on the keyboard in dismay, lol.

    2800 x 2100 - 1M
    480 x 600 - 113K
    680 x 512 - 237K
    640 x 400 - 22K
    1920 x 960 - 276K
    1616 x 1080 - 1M
    700 x 840 - 278K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited March 7

    Just when I thought I could start scribbling on screencaps, and having some fun trying to make a fun render, nope.

    Just like Cooler testing madness (GN YT vid), there are so many things this could be from the hair to the figure shape combo, to something glitching during the conversion process, to just windows10 being windows10. lol.

    Looks like I need to go back, and just try the conversion, yet again, ugh. This may take a smidgen longer than I had hoped. Things were sort of going smooth until this morning with the exception of forgetting a step or few from RiverSoft's vid (RiverSoftArt YT vid on Character Converter).

    Time for some coffee. at least, I have a lot to scribble on and try to condense into a coherent post. Lots of fun. Oh, congrats to Isaac Arthur for the Space Pioneer Award (NSS webpage).


    ok, looks like it may be something with that hair specifically, I just scrapped all the gogo dsf files, converted it again, and still did it, then tried Eve and got the same result with that hair. tried another on-shoulder hairstyle, no probs. hmmm. FYI, both styles are by the same PA, so I don't think it's a difference in workflow between them, don't think. Happenstance appears to be ok, no clue whats going on with Innocent Hair. I do recall Happenstance can be slow to complete rendering with some lights, tho not as bad as some other styles I've endured.

    Note to self, make sure gen8 doesn't already have a figure named what your converting. In my haste to look at the bone opacity anomaly and converting FWSA Phoebe for a quick test, I had fogoten that Raziel-Jessaii Phoebe existed on gen8.

    so it looks like 'zero figure' sets them dials to zero, and the script converts that setting over to the gen8 dial. it apparently isn't saved, cuz clearing the scene is all it takes for the bone opacity dials to go back to normal. so I'm not sure if the 'zero figure' was a good choice for doing zero shape and zero pose faster now.

    1920 x 1050 - 1M
    1920 x 1050 - 1M
    900 x 460 - 115K
    3000 x 1025 - 2M
    1920 x 1080 - 300K
    1680 x 1360 - 761K
    544 x 567 - 29K
    1080 x 540 - 55K
    554 x 523 - 27K
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited March 9

    ok, after many cups of coffee, having a small chat with RiverSoft about some 'glitches' and considering how the script does not save the mistaken conversions for some reason, I have decided to stick with mu original thought of "zero out EVERYTHING" between conversions of figure sections.

    again, bone opacity apparently isn't saved, cuz clearing the scene is all it takes for the bone opacity dials to go back to normal.

    Right is the original GoGo for Mei Lin6 (Mei Lin swapped out with Aiko6 and Aiko8 respectively) wearing After School 2 with Discipline for After School 2 textures (I attempted to d-force the skirt, that went kind of eh), and The Jogger sneakers.  Left is Gogo converted to gen8 using that Gen6-to-gen8 script, wearing the Pearl outfit from FWSA Mable auto flopped on Gen8 with the shinies shaders, Frolic stockings, and cool Amazing Morphing Boots.

    I will admit my attempt of mimicking Gogo's makeup in Gimp using CS Grace's maps wasn't the best given my chicken scratches skills, lol.

    so, maps aside, I find it best to convert just the head dial, then zero EVERYTHING out and then convert the body. because often a head or body shape will use another dial, and the script ends up lumping that in with the control dial for the full figure rather than just the head or body, if its all converted at once. I also find it best to do the ear separately as well for elves/fae, with the understanding it will need some work in hexagon.

    Left is the original FW Eve, right is the conversion using ADSI Sewa's maps. Not quite as similar to Eve's freckles I will admit, maybe some other maps would do better like Madeline or Autumn, maybe.

    I kind of expected the ears to fail the instant I had seen just how defined the point was on them for FW Eve.

    I could probably smooth that over in hex with an hour or so of poking at mesh points. Some time ago I had attempted the transfer utility thing to migrate the ear shape I had made from G3F over to G3M, and even that needed major work on effectively identical mesh, so I'm not surprised at all with elf ears transferring rather poorly. The rest of the conversion went really well I think.

    So in closing as I look for a cool outfit combo for DES Cassia. Watch RiverSoft's YT vid a couple of times, make backups of and be prepared to delete the converted DSF files and try a couple of times before running with a converted character for scenes and stuff.

    2800 x 2800 - 1M
    1300 x 419 - 648K
    1920 x 1050 - 403K
    2800 x 2800 - 980K
    1920 x 1050 - 1M
    3840 x 1050 - 2M
    Post edited by ZarconDeeGrissom on
  • ZarconDeeGrissomZarconDeeGrissom Posts: 5,358
    edited 7:52AM

    ok, pushing this ahead with a teaser as I'm still scribbling on notes and screencaps for the other stuff. still alive, the past few weeks have been shall we say interesting.

    That is Tigerlily converted to gen8 using that nifty gen6 to gen8 shape converter thing with FWSA Paloma's maps. I did eventually change the maps earlier this week to something else that I'm still scribbling notes for. Overall not that bad I think. in the meantime, going back to 2018 when I started looking at designing a new workstation case for the first time since I made the one I have now back in 2004

    I think this GamersNexus vid (YouTube vid) was partly my fault after poking and prodding at larger than ATX cases, lol.

    And exasperated by the XL-ATX vs E-ATX threadriper3000 boards last year.

    I had no idea just how messed up things were, I thought EEB or E-ATX was and always was the same thing, 12x13 inches. Apparently, E-ATX is like looking for a case that has top-mounted radiator support. Just because it sais a 240mm rad will fit, does not imply you can put a motherboard in there with the radiator, lol.

    1400 x 1400 - 283K
    1400 x 734 - 847K
    480 x 408 - 30K
    Post edited by ZarconDeeGrissom at
Sign In or Register to comment.