Loading...
> >

Algovincian Non-Photorealistic Rendering (NPR) 2019/2020

12346

Comments

  • towdow3towdow3 Posts: 83

    How did you build the algorithm? I have been struggling with getting decent linework from my renders. I dont wanna use zbrush or blender to get them.

  • 3Diva3Diva Posts: 11,523
    algovincian said:
    JeremyD said:
    algovincian said:
    Divamakeup said:

    I would love to do a comic, but I'm not sure if I have the writing skills and/or visual storytelling skills for a decent one. So I'm really just looking for a few assets to maybe do a short 4 or 5 page thing to see if I even have what it takes to put something like that together. I know it can be some work going from a story, to storyboarding each scene, to getting everything to flow right visually. Though being stuck at home right now because of Covid-19 is giving me no excuses not to buckle down and at least try it. As long as I can get the assets I need together to do it.

    Yeah I know exactly how you feel. It's a totally different skillset. Putting together the actual storytelling & storyboarding seems daunting to me, too. Collecting/creating all of the necessary assets and handling the actual NPR is challenging/time consuming enough. Even if I had unlimited time (which I don't), not sure it's in my wheelhouse lol.

    I'd still like to take a stab at doing some finished pages to keep the development moving forward. Would be helpful to use as something concrete to potentially present to storytellers who may wish to collaberate in the future, too.

    In the mean time, here's a few more background environments - thank you @Stonemason! Ran out of time today, but I'll throw some figures and text on top in the next couple of days (and hold off on judging the BG algo's output until then):

    - Greg

     

     

    omgosh these are incredible :0 especially the last one. now im in the mood to watch some studio ghibli. 

     

    Thanks @JeremyD - always nice to hear when people enjoy them! I googled Studio Ghibli and found that they've created many animations. I'd like to check some out - do you have any favorites/suggestions?

     

    mindsong said:
    JeremyD said:
    algovincian said:
    Divamakeup said:

    I would love to do a comic, but I'm not sure if I have the writing skills and/or visual storytelling skills for a decent one. So I'm really just looking for a few assets to maybe do a short 4 or 5 page thing to see if I even have what it takes to put something like that together. I know it can be some work going from a story, to storyboarding each scene, to getting everything to flow right visually. Though being stuck at home right now because of Covid-19 is giving me no excuses not to buckle down and at least try it. As long as I can get the assets I need together to do it.

    Yeah I know exactly how you feel. It's a totally different skillset. Putting together the actual storytelling & storyboarding seems daunting to me, too. Collecting/creating all of the necessary assets and handling the actual NPR is challenging/time consuming enough. Even if I had unlimited time (which I don't), not sure it's in my wheelhouse lol.

    I'd still like to take a stab at doing some finished pages to keep the development moving forward. Would be helpful to use as something concrete to potentially present to storytellers who may wish to collaberate in the future, too.

    In the mean time, here's a few more background environments - thank you @Stonemason! Ran out of time today, but I'll throw some figures and text on top in the next couple of days (and hold off on judging the BG algo's output until then):

    - Greg

     

     

    omgosh these are incredible :0 especially the last one. now im in the mood to watch some studio ghibli. 

    absolute magic going on here.

    --ms

     

    I don't know about magic, @mindsong - just the result of a process that's been evolving for almost 3 decades now. Anyway, thanks for taking the time to look and comment. It's much appreciated.

     

    Divamakeup said:
    algovincian said:

    I like the more detailed one better, too, when looking at it on a computer monitor and it gets displayed pixel perfect (no resizing). But have you looked at it on a phone? When it's smaller and being resized, bad things happen to fine lines.

    Here's a simple, concrete example illustrating what I'm talking about:

    The top row is a cropped portion of the less detailed image, with the original 128x128 pixel version in the middle, and resized versions to either side. The bottom row is the same thing for the more detailed image.

    To me, the fine lines do not scale well. I understand that different resizing algorithms have different strengths and weaknesses, but any resizing will still always cause problems. I'm exploring the less detailed style because display on mobile devices has always been a consideration for me.

    Consider the simplest case: a 1 pixel wide black line. If you want to scale it down to 50%, how do you do it? Inevitably, it becomes a grey line. Also, how do you scale it up 150%? Same thing happens - the introduction of greys. This has the effect of making lines fatter & fuzzier. It's an unfortunate result of digital sampling and cannot be avoided. 

    It's worth noting that it's not only scaling down that causes an issue. Scaling up does as well. The fact that increasing resolution (more pixels) can actually hurt image fidelity can be a bit counter intuitive, but true none the less.

    - Greg

    ETA: I think maybe a new thread for finished comic pages is a good idea - if I ever actually finish some ;)

    Oh yeah - I see what you mean. I looked at the thread on my phone and you're right, the first does look a bit better on such a small screen. Are your Algos able to do something that is a little bit in between the two - with the thicker linework of the first image but maybe just a tad more details? Almost like a mix of the two but leaning more toward the first one? 

     

    Funny you should say that, as what you're describing sounds an awful lot like the process I currently go through. As you know, I save close to 1GB of image data every time I run the algos. This includes multiple styles, intermediate images, different masks generated based on the analysis passes, etc. The table of images below represents just a few of these images:

    The left column is low resolution output from the algos, while the right column is higher resolution. This is just one of the many fractal-ish concepts employed by the algos. They're all created based on the same analysis passes, but as you can see, it provides for many variations.

    The "Final" images were created from the Alpha (which is simplified and has thicker lines), the Color (which was the "Final" full color output earlier in 2019), and the BW. I also included the Z-Depth and Detail masks as an example of analysis passes and derivative images used in the processing of the final output.

    As always, it's a work in progress (I know you know exactly what I mean from comments you've made about your own work).

    - Greg

    ETA: These images are cropped versions of the Bone Breaker I posted earlier in the thread. They've all been resized to the same small size so they'd be easier to compare. It should be noted that had they been larger versions, it would have substantially changed the look of each. This is an unavoidable consequence of digital sampling and the reason for the discussion about it a few posts back.

    That is too cool! I really like the Alpha passes! I'm looking forward to seeing more. laugh

  • algovincianalgovincian Posts: 2,613
    towdow3 said:

    How did you build the algorithm? I have been struggling with getting decent linework from my renders. I dont wanna use zbrush or blender to get them.

    The lines were not rendered in DS. Custom shaders were written for 3DL in DS to render out 8 different (what I refer to as) analysis passes. These allow custom fractal algorithms to understand the scene and make strokes. The decisions about exactly what strokes the algorithms make are governed by neural nets.

    There's more too it than that (simplification algorithms, smoothing algorithms, HDR algorithms, etc.), but those are the broad brush strokes. At the heart of the whole process is the integration of all of these and other systems via scripting/automation. It's the automation that makes everything possible since it's so convoluted.

    - Greg

  • Xandyr78Xandyr78 Posts: 117

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

  • algovincianalgovincian Posts: 2,613
    Xandyr78 said:

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

    Thanks @Xandyr78. It's always nice to hear that people enjoy the images rendered by the algos - much appreciated!

    Been kind of MIA for a while, so figured it might be fun to play around a bit today using Spidey to test a lithe/spry custom morph made for g8m a while back. It's a rather extreme morph, so it seemed like a good idea to test it with some challenging poses after rigging with @Zev0 and @Discobob's excellent Shape Rigger Plus (https://www.daz3d.com/shape-rigger-plus):

    Still working on depicting figures at different distances/levels of detail, paying particular attention to how their line weights compare with each other as well as with the balloon/text. Also working on the way the figures are rendered compared to the background (trying to create enough separation while maintaining some overall continuity).

    - Greg

     

     

     

  • barbultbarbult Posts: 24,244

    I like the way Spiderman #2 is clinging to the edge of the frame and the bottom Spiderman is at the bottom edge of the frame. It adds a bit of tension to the scene.

  • 3Diva3Diva Posts: 11,523
    edited July 2020
    algovincian said:
    Xandyr78 said:

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

    Thanks @Xandyr78. It's always nice to hear that people enjoy the images rendered by the algos - much appreciated!

    Been kind of MIA for a while, so figured it might be fun to play around a bit today using Spidey to test a lithe/spry custom morph made for g8m a while back. It's a rather extreme morph, so it seemed like a good idea to test it with some challenging poses after rigging with @Zev0 and @Discobob's excellent Shape Rigger Plus (https://www.daz3d.com/shape-rigger-plus):

    Still working on depicting figures at different distances/levels of detail, paying particular attention to how their line weights compare with each other as well as with the balloon/text. Also working on the way the figures are rendered compared to the background (trying to create enough separation while maintaining some overall continuity).

    - Greg

     

     

     

    I LOVE the morph! The over-exaggerated super slim limbs and large hands and feet remind me of my own character design: https://www.daz3d.com/forums/discussion/comment/5751341/#Comment_5751341

    That's a stylized look I really like. Very cool, Greg! 

    The background line weights could be a tad thinner, imo - but you're doing wonderful work with making the foreground figure pop. The slightly lighter background helps with that as well. Excellent work, Greg!

    Post edited by 3Diva on
  • algovincianalgovincian Posts: 2,613
    edited March 2021
    barbult said:

    I like the way Spiderman #2 is clinging to the edge of the frame and the bottom Spiderman is at the bottom edge of the frame. It adds a bit of tension to the scene.

    Thanks, @Barbult. Having the figures interact with the image borders has always been part of the style that I've been trying to evolve. It's usually the feet glued the bottom with an extremely low perspective for the background, but it's fun to use the walls, too. IIRC, first time I tried it was actually an Iray render:

    I remember struggling to find something to do with his foot to make it look like it was interacting with the HDRI (like his hand looks). I finally gave up and just tried to make it look like his leg was all bunched up like that from pushing against the image border.

     

     

    Divamakeup said:
    algovincian said:
    Xandyr78 said:

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

    Thanks @Xandyr78. It's always nice to hear that people enjoy the images rendered by the algos - much appreciated!

    Been kind of MIA for a while, so figured it might be fun to play around a bit today using Spidey to test a lithe/spry custom morph made for g8m a while back. It's a rather extreme morph, so it seemed like a good idea to test it with some challenging poses after rigging with @Zev0 and @Discobob's excellent Shape Rigger Plus (https://www.daz3d.com/shape-rigger-plus):

    Still working on depicting figures at different distances/levels of detail, paying particular attention to how their line weights compare with each other as well as with the balloon/text. Also working on the way the figures are rendered compared to the background (trying to create enough separation while maintaining some overall continuity).

    - Greg

     

     

     

    I LOVE the morph! The over-exaggerated super slim limbs and large hands and feet remind me of my own character design: https://www.daz3d.com/forums/discussion/comment/5751341/#Comment_5751341

    That's a stylized look I really like. Very cool, Greg! 

    The background line weights could be a tad thinner, imo - but you're doing wonderful work with making the foreground figure pop. The slightly lighter background helps with that as well. Excellent work, Greg!

    I agree with you, Diva, although it doesn't bother me as much on my phone as it does my monitor. The thought was to render the background in less detail, and because of the way the algos function, this is done by rendering at lower resolution (as was discussed in a few posts on the previous page). As a consequence, the lines also got thicker.

    Below is what it looks like with the BG rendered at a higher res/thinner lines. Still worried about the higher res not resizing well (which is a bigger concern on mobile devices):

    On another note, it's amazing how different things look with fresh eyes. Looking at the image again, it looks like there's a budding 12 year old girl wearing the suit lol. Not exactly what I was going for!

    The morph itself doesn't look like that, so it must be a combination of the way the suit is conforming, the way the algos drew it, and the way the lighting is accentuating it:

    The morph is part of a series of caricature morphs I've been working on. Hopefully there will end up being a half dozen or so. The thought is to design them so that when dialed in to existing characters, the general shape is changed, but the recognizable features of the character aren't lost. Here's another one of the extreme caricature shapes - a clownish type:

    I remember your posts in that other comic thread. I really like the body shape of your character, and enjoyed the 2 character sheets you did. Have you had a chance to do any more?

    - Greg

    Post edited by algovincian on
  • barbultbarbult Posts: 24,244

    I didn't notice the "budding" at all until you pointed it out.

  • algovincianalgovincian Posts: 2,613

    This time the entire image was rendered from a complete 3D scene that included both the foreground and background - unlike the Spidey image where the figures and BG were all rendered separately and composited.

    Both approaches have their advantages. In this case, the transition from foreground to background is made gradually by the wires and such at the top of the image. Seems to work nicely, I think. I've also got some ideas involving parallax effects which are easier done with elements rendered seperately, though.

    The monochromatic palette is appealing, but the lack of full color information can make it harder to get easily discernible detail. Several techniques were used in this case, like using the z-depth to control exposure and DOF. The use of DOF on lineart has always interested me, though the jury is still out.

    - Greg

  • algovincianalgovincian Posts: 2,613
    edited August 2020

    Decided to re-run one of the first images I posted when I started the original Non-Photorealistic Renders (NPR) thread here in the DAZ forums way back in April of 2015. There are still things I like about the original algos more than the current, and I have to keep reminding myself that changes were made along the way for a reason (mainly new styles, and making them work for the general case much better now). 

    Anyway, same style as the last, only the DOF lens blur was replaced with a smoothing algorithm dynamically controlled by the z-depth:

    Oh, and the necessary evil (limitation) of the shader based z-depth that supports transparency rears its ugly head once again! It really stands out on the guy in the front, where the tree behind his hair is rendered as being closer to the viewer because it's seen through the hair . . . sigh. His fine whispy white hair was rendered weird anyway - don't know how to fix either problem lol.

    - Greg

     

    Post edited by algovincian on
  • 3Diva3Diva Posts: 11,523
    algovincian said:

    This time the entire image was rendered from a complete 3D scene that included both the foreground and background - unlike the Spidey image where the figures and BG were all rendered separately and composited.

    Both approaches have their advantages. In this case, the transition from foreground to background is made gradually by the wires and such at the top of the image. Seems to work nicely, I think. I've also got some ideas involving parallax effects which are easier done with elements rendered seperately, though.

    The monochromatic palette is appealing, but the lack of full color information can make it harder to get easily discernible detail. Several techniques were used in this case, like using the z-depth to control exposure and DOF. The use of DOF on lineart has always interested me, though the jury is still out.

    - Greg

    WOW! I REALLY Like This! While I think it could benefit from some color - the DOF effect really helps the foreground pop. I think this is awesome!

  • 3Diva3Diva Posts: 11,523
    edited August 2020
    algovincian said:

    Decided to re-run one of the first images I posted when I started the original Non-Photorealistic Renders (NPR) thread here in the DAZ forums way back in April of 2015. There are still things I like about the original algos more than the current, and I have to keep reminding myself that changes were made along the way for a reason (mainly new styles, and making them work for the general case much better now). 

    Anyway, same style as the last, only the DOF lens blur was replaced with a smoothing algorithm dynamically controlled by the z-depth:

    Oh, and the necessary evil (limitation) of the shader based z-depth that supports transparency rears its ugly head once again! It really stands out on the guy in the front, where the tree behind his hair is rendered as being closer to the viewer because it's seen through the hair . . . sigh. His fine whispy white hair was rendered weird anyway - don't know how to fix either problem lol.

    - Greg

     

    The blurred background gives it an almost watercolor or ink wash look. It makes it look even more like "traditional" art, I think. Nice! 

    I think the hair you mentioned is a bit too "whispy" and doesn't feel like it fits with the style of the rest of the image. The tree poking through it, like you pointed out, is also an unfortunate side effect of the wispiness. Maybe bringing up the "cutout opacity" a lot on the hair would work to help eliminate both issues? It would help the algos have more to work with and, hopefully, give some nice lines there and also maybe help the algos to separate the hair from the tree? Unless it's strand-based hair, in which case perhaps increasing the strand width of the hair would work?

    All in all that style really is quite cool! It's a fantastic hand-inked look that works great and looks really good!

    (That text made me laugh! NICE!)

    Post edited by 3Diva on
  • algovincianalgovincian Posts: 2,613

    Like the new name! Is @3Diva the name you'll be publishing your comic under? Even though I don't post in them all that much, I've been lurking in the comic threads, and it seems as though you're getting more serious - great news!

    One thing that was discussed is consistency, which is very important to me. I'd like to avoid getting halfway through illustrating a story only to realize that some situations are not possible (read: doesn't look good) to render in the same style, if that makes sense. This is one reason why I run so many tests.

    Also, I can't remember which one of those comic threads it was in, but at some point there was a discussion about the importance of, or lack thereof, the art style. It was suggested that the art doesn't really matter as long as the story telling is good. Personally, I couldn't disagree more. There's such a flood of content out there, I feel you've got to stand out. It's going to be the art that initially gets somebody to do any reading at all initially. Both need to be up to par.

    Anyway, the cemetery scene was setup back in 2015 and nothing was changed. Just loaded the old scene, rendered out the analysis passes, and ran the current algos. The guy in the foreground is actually an HD Genesis figure (perhaps one of the first):

    https://www.daz3d.com/walker-hd-for-genesis

    The hair came with him and was intentionally left alone to see what would happen. Sort of figured there might be issues and that turned out to be the case lol. The obvious solution is to change the hair, but there's part of me that wants to try to get the algos to render it (decently) as is. I'll keep playing with it.

    As far as color goes, the more I look at the Spiderman test, the more I don't like it. I agree that color would be nice, and I've been working on adding color to the latest style variation. It's not there yet. It's definitely more tricky to get it right for color, and especially so for the general case as opposed to a specific image. I'm currently running a large batch over & over as I tweak the new color algo, checking how the tweaks affect each as I go.

    Here's the current state of this new color style for the last 2 images:

    The reactor looks ok, but it's still hit or miss in many other cases (not a huge fan of how the cemetery was rendered in this style). In general, I do like the monochromatic style and I'm not totally against a full color cover with a monochromatic style on the inside . . . maybe? I dunno.

    - Greg

  • 3Diva3Diva Posts: 11,523
    edited August 2020
    algovincian said:

    Like the new name! Is @3Diva the name you'll be publishing your comic under?

    Thank you! I did the name change because I realized that "Divamakeup" as a vendor name could be pretty misleading since I rarely do any products that are makeup related. I wanted to keep the "Diva", since a lot of people just call me "Diva" and I've gotten so used to it, but I wanted to incorporate the 3D aspect as well - I know it's probably overdone, but I think incorporating the "3D" into the name is a nice way to pay homage to this important art form. :D 

    I'm not sure if I'll be publishing my comic under my legal name or my PA name. I haven't decided on that. Since I'm going to be going with an NPR look, I want to be able to draw people in without making it obvious (at first) that it's 3D. I know that in the comic community there is a real stigma against 3D art and 3D comics in particular. I want people to give the book a chance without them bring in their prejudices against 3D art. At the same time, if people find out that it's 3D after they start liking it (hopefully) then maybe that can change their minds about 3D art in comics. I'm sure, no matter how much I "NPR it", that people really familiar with 3D will know it's 3D, but those who aren't that familiar with it or who haven't really given 3D much attention, I don't think they would realize that it's 3D. I don't know, it's a tough decision on how to present it. Using the name 3Diva would likely make it too obvious that it's 3D - if that's something I didn't want to be known up-front until people get into the comic or give it a chance first, I don't know... 

     

    algovincian said:

    One thing that was discussed is consistency, which is very important to me. I'd like to avoid getting halfway through illustrating a story only to realize that some situations are not possible (read: doesn't look good) to render in the same style, if that makes sense. This is one reason why I run so many tests.

    Also, I can't remember which one of those comic threads it was in, but at some point there was a discussion about the importance of, or lack thereof, the art style. It was suggested that the art doesn't really matter as long as the story telling is good. Personally, I couldn't disagree more. There's such a flood of content out there, I feel you've got to stand out. It's going to be the art that initially gets somebody to do any reading at all initially. Both need to be up to par.

    Yeah, I do need to run more "tests" as well. And just start posting different images done in the style that I want to go with. That way I can get feedback on it and make sure that it "works" in a variety of different scenes. So far, though, I do believe I have a style nailed down that I REALLY LIKE - I've tested it quite a bit and am super happy with the way it looks, but you're right, I should do even more tests and post the results to get feedback.

     

     

    algovincian said:

    Anyway, the cemetery scene was setup back in 2015 and nothing was changed. Just loaded the old scene, rendered out the analysis passes, and ran the current algos. The guy in the foreground is actually an HD Genesis figure (perhaps one of the first):

    https://www.daz3d.com/walker-hd-for-genesis

    The hair came with him and was intentionally left alone to see what would happen. Sort of figured there might be issues and that turned out to be the case lol. The obvious solution is to change the hair, but there's part of me that wants to try to get the algos to render it (decently) as is. I'll keep playing with it.

    As far as color goes, the more I look at the Spiderman test, the more I don't like it. I agree that color would be nice, and I've been working on adding color to the latest style variation. It's not there yet. It's definitely more tricky to get it right for color, and especially so for the general case as opposed to a specific image. I'm currently running a large batch over & over as I tweak the new color algo, checking how the tweaks affect each as I go.

    Here's the current state of this new color style for the last 2 images:

    The reactor looks ok, but it's still hit or miss in many other cases (not a huge fan of how the cemetery was rendered in this style). In general, I do like the monochromatic style and I'm not totally against a full color cover with a monochromatic style on the inside . . . maybe? I dunno.

    - Greg

    I do like the reactor with color - though it looks pretty good in black and white too. I think, because it's sci-fi, you could really make the colors "pop" there if you wanted. For the cemetery scene, I think you're right - the monochromatic style works best for it. Color can be "comforting", and monochromatic can be a bit unsettling sometimes, that's one of the reasons I think that it works so well with horror. Color makes it easier to quickly identify each individual figure and what's going on, with the monochromatic style it's a bit less clear at a glance about what all you're seeing, which is fantastic for the genre of horror, imo.

    Post edited by 3Diva on
  • algovincianalgovincian Posts: 2,613
    edited August 2020

    Saw @Smay's Liquid Armor in another thread last week:

    https://www.daz3d.com/liquid-armor-for-genesis-8-females

    So I picked it up just to do a Venom-esque panel for fun (thanks Diva). Thought of the stripes on Aureska and ended up using a modified version of her mats:

    https://www.daz3d.com/aureska-hol-hd-for-zelara-8

    The tongue was something I made years ago when learning to use the figure setup tools in DS (really easy to make a quick chain type rigging).

    I've always liked the idea of a panel being a window that figures could step out of. After playing with her pose and getting one foot out of the confines of the panel, I remembered Them Bones was part of my library and thought a pile might look good outside the panel, too:

    https://www.daz3d.com/them-bones-legacy-bundle-with-iray-updates

    Used a BG that I've already rendered - one of my favorites from way back when called The Ministry:

    https://www.daz3d.com/the-ministry

    Put it all together and in the course of an evening this was the end result:

    Having assets that I could just grab from my library and modify is a satisfying and efficient process. The whole thing was so much fun to create and reminded me exactly why I enjoy DS so much. And now I have this figure all setup to use in the future, too.

    - Greg

     

    Post edited by algovincian on
  • algovincianalgovincian Posts: 2,613
    3Diva said:
    algovincian said:

    Like the new name! Is @3Diva the name you'll be publishing your comic under?

    Thank you! I did the name change because I realized that "Divamakeup" as a vendor name could be pretty misleading since I rarely do any products that are makeup related. I wanted to keep the "Diva", since a lot of people just call me "Diva" and I've gotten so used to it, but I wanted to incorporate the 3D aspect as well - I know it's probably overdone, but I think incorporating the "3D" into the name is a nice way to pay homage to this important art form. :D 

    I'm not sure if I'll be publishing my comic under my legal name or my PA name. I haven't decided on that. Since I'm going to be going with an NPR look, I want to be able to draw people in without making it obvious (at first) that it's 3D. I know that in the comic community there is a real stigma against 3D art and 3D comics in particular. I want people to give the book a chance without them bring in their prejudices against 3D art. At the same time, if people find out that it's 3D after they start liking it (hopefully) then maybe that can change their minds about 3D art in comics. I'm sure, no matter how much I "NPR it", that people really familiar with 3D will know it's 3D, but those who aren't that familiar with it or who haven't really given 3D much attention, I don't think they would realize that it's 3D. I don't know, it's a tough decision on how to present it. Using the name 3Diva would likely make it too obvious that it's 3D - if that's something I didn't want to be known up-front until people get into the comic or give it a chance first, I don't know... 

    Too bad the domain (.com) is already taken as it's a nice short one. It can be yours for just $2,995, though!

    I know exactly what you mean about the negative 3D association as far as comics are concerned. There is so much time to be saved producing a comic using 3D in the workflow! This is assuming, of course, desirable NPR. And this is why we are on this quest ;)

    - Greg

  • 3Diva3Diva Posts: 11,523
    algovincian said:

    Saw @Smay's Liquid Armor in another thread last week:

    https://www.daz3d.com/liquid-armor-for-genesis-8-females

    So I picked it up just to do a Venom-esque panel for fun (thanks Diva). Thought of the stripes on Aureska and ended up using a modified version of her mats:

    hahah You're welcome! I really like that outfit, it's very cool! It's like a mix of Venom and Witchblade! Smay makes some really neat stuff, imo. :) I'm glad you're having fun with it. The figure you rendered is SO CREEPY! lol

  • 3Diva3Diva Posts: 11,523
    edited August 2020
    algovincian said:
    3Diva said:
    algovincian said:

    Like the new name! Is @3Diva the name you'll be publishing your comic under?

    Thank you! I did the name change because I realized that "Divamakeup" as a vendor name could be pretty misleading since I rarely do any products that are makeup related. I wanted to keep the "Diva", since a lot of people just call me "Diva" and I've gotten so used to it, but I wanted to incorporate the 3D aspect as well - I know it's probably overdone, but I think incorporating the "3D" into the name is a nice way to pay homage to this important art form. :D 

    I'm not sure if I'll be publishing my comic under my legal name or my PA name. I haven't decided on that. Since I'm going to be going with an NPR look, I want to be able to draw people in without making it obvious (at first) that it's 3D. I know that in the comic community there is a real stigma against 3D art and 3D comics in particular. I want people to give the book a chance without them bring in their prejudices against 3D art. At the same time, if people find out that it's 3D after they start liking it (hopefully) then maybe that can change their minds about 3D art in comics. I'm sure, no matter how much I "NPR it", that people really familiar with 3D will know it's 3D, but those who aren't that familiar with it or who haven't really given 3D much attention, I don't think they would realize that it's 3D. I don't know, it's a tough decision on how to present it. Using the name 3Diva would likely make it too obvious that it's 3D - if that's something I didn't want to be known up-front until people get into the comic or give it a chance first, I don't know... 

    Too bad the domain (.com) is already taken as it's a nice short one. It can be yours for just $2,995, though!

     

    $2,995! lol Wow... That's not happening. :P 

     

    algovincian said:
    3Diva said:
    algovincian said:

    Like the new name! Is @3Diva the name you'll be publishing your comic under?

    Thank you! I did the name change because I realized that "Divamakeup" as a vendor name could be pretty misleading since I rarely do any products that are makeup related. I wanted to keep the "Diva", since a lot of people just call me "Diva" and I've gotten so used to it, but I wanted to incorporate the 3D aspect as well - I know it's probably overdone, but I think incorporating the "3D" into the name is a nice way to pay homage to this important art form. :D 

    I'm not sure if I'll be publishing my comic under my legal name or my PA name. I haven't decided on that. Since I'm going to be going with an NPR look, I want to be able to draw people in without making it obvious (at first) that it's 3D. I know that in the comic community there is a real stigma against 3D art and 3D comics in particular. I want people to give the book a chance without them bring in their prejudices against 3D art. At the same time, if people find out that it's 3D after they start liking it (hopefully) then maybe that can change their minds about 3D art in comics. I'm sure, no matter how much I "NPR it", that people really familiar with 3D will know it's 3D, but those who aren't that familiar with it or who haven't really given 3D much attention, I don't think they would realize that it's 3D. I don't know, it's a tough decision on how to present it. Using the name 3Diva would likely make it too obvious that it's 3D - if that's something I didn't want to be known up-front until people get into the comic or give it a chance first, I don't know... 

     

    This is assuming, of course, desirable NPR. And this is why we are on this quest ;)

     

    For sure! I think it's a fun quest too. :)

    I think some comics are doing pretty well without NPR, though. I see Linwelly's comic and Drew's comic and they seem to be doing well without an NPR style. I'm glad that there is room for a wide range of 3D comic styles. :D I think for me, personally, I do want to go the NPR route - maybe it's my work in traditional comics, but I've always really enjoyed most the styles that have the look of being hand done. I do enjoy seeing a wide variety of styles though - from Photorealism to styles like yours that are pretty much indistinguishable from hand-drawn work. It's really fun to see the many forms that 3D comics can take.

    Post edited by 3Diva on
  • algovincianalgovincian Posts: 2,613
    edited March 2021

    Nothing quite like going all necro on your own thread . . .

    Anyway, just made a simple animated GIF and wanted to test to see if it actually animates after uploading to the forums:

    - Greg

    Post edited by algovincian on
  • algovincianalgovincian Posts: 2,613
    edited March 2021

    Doh! Guess it doesn't work when uploading here to the forums. Let me try uploading to another domain and referencing it:

    - Greg

    Post edited by algovincian on
  • algovincianalgovincian Posts: 2,613
    edited March 2021

    It worked . . . woot! Looks pretty good given that the file size on this little slideshow is only 345KB.

    The figures were Mavka and Gosha - both creations by @smay. They look fantastic rendered as NPR. If you're out there @smay and see this, please make more ;)

    - Greg

    Post edited by algovincian on
  • Sven DullahSven Dullah Posts: 7,621

    algovincian said:

    Doh! Guess it doesn't work when uploading here to the forums. Let me try uploading to another domain and referencing it:

    - Greg

    Excellent work as alwaysyessmiley

  • juvesatrianijuvesatriani Posts: 556

    algovincian said:

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

    Thanks @Xandyr78. It's always nice to hear that people enjoy the images rendered by the algos - much appreciated!

    Been kind of MIA for a while, so figured it might be fun to play around a bit today using Spidey to test a lithe/spry custom morph made for g8m a while back. It's a rather extreme morph, so it seemed like a good idea to test it with some challenging poses after rigging with @Zev0 and @Discobob's excellent Shape Rigger Plus (https://www.daz3d.com/shape-rigger-plus):

    Still working on depicting figures at different distances/levels of detail, paying particular attention to how their line weights compare with each other as well as with the balloon/text. Also working on the way the figures are rendered compared to the background (trying to create enough separation while maintaining some overall continuity).

    - Greg

     

     

     

    Wow !! this is very impressive style , really hope your tools will be in store someday or at least you can run web services for people who want NPR render with their scenes

     

  • algovincianalgovincian Posts: 2,613

    Sven Dullah said:

    algovincian said:

    Doh! Guess it doesn't work when uploading here to the forums. Let me try uploading to another domain and referencing it:

    - Greg

    Excellent work as alwaysyessmiley

     

    Thanks Sven - working with Genesis 1 again was fun. I think the hair and clothes on Mavka rendered great and fit nicely with the stand-alone figure/clothes/hair of Gosha. Setting up these renders was the most fun I've had in DS in quite some time!

     

    juvesatriani said:

    algovincian said:

    Every time I see a new piece from you, Algovincian, it makes me long for the day when this process will be something available to the rest of us lowly paeons.  Incredible work!

    Thanks @Xandyr78. It's always nice to hear that people enjoy the images rendered by the algos - much appreciated!

    Been kind of MIA for a while, so figured it might be fun to play around a bit today using Spidey to test a lithe/spry custom morph made for g8m a while back. It's a rather extreme morph, so it seemed like a good idea to test it with some challenging poses after rigging with @Zev0 and @Discobob's excellent Shape Rigger Plus (https://www.daz3d.com/shape-rigger-plus):

    Still working on depicting figures at different distances/levels of detail, paying particular attention to how their line weights compare with each other as well as with the balloon/text. Also working on the way the figures are rendered compared to the background (trying to create enough separation while maintaining some overall continuity).

    - Greg

     

     

     

    Wow !! this is very impressive style , really hope your tools will be in store someday or at least you can run web services for people who want NPR render with their scenes

     

    I do, too @juvesatriani. This batch of renders will be added to the database I'm creating for supervised training. The autonomous way that these NPR images are rendered allows me to create a large data set. I believe (hope) that the consistency of the process and creation of the training pairs will lead to a successful network design.

    There are still some major hurdles to be overcome like memory concerns and the design of certain layers of the network (especially the input), but progress is being made. Hopefuly, I'll finish it before I die!

    - Greg

  • Xandyr78Xandyr78 Posts: 117

    Hey, Algo-master!  Just checking in to ask about progress.  That Spidey image looks AMAZING.

  • algovincianalgovincian Posts: 2,613

    Just saw this message - sorry for not getting back sooner. I'm still working on designing/training a convolutional neural network to replace the current convoluted process.

    Is it possible? I don't know. That's a good question.

    Am I smart enough to get it to work? That's an even better question! lol

    - Greg

  • Worlds_EdgeWorlds_Edge Posts: 2,152

    I would never have guessed that the spidey images are renders.  Wow!  Great work.

  • Xandyr78Xandyr78 Posts: 117

    algovincian said:

    Just saw this message - sorry for not getting back sooner. I'm still working on designing/training a convolutional neural network to replace the current convoluted process.

    Is it possible? I don't know. That's a good question.

    Am I smart enough to get it to work? That's an even better question! lol

    - Greg

    Well, if you ever need/want lowly artists to help you test this thing (I am no code-monkey, but I'm a graphic designer, so the art side I've got), I'd love to try and put it through its paces.  This would be AMAZING for a graphic novel I've wanted to create for a very long time. 

  • Llola LaneLlola Lane Posts: 9,344

    Just found this thread (and the numeous others algovincian)  I've spend a good part of an hour reading through them.  Looks like I'm late to the party... Seems like this thread has hit a stand still :(... But I am enjoying looking at all the NPR artworks.  Soooo many interpretations... Not sure I understand the algorithm stuff... but it has gotten me experimenting with some of the features in my PSP7 ;)  Thanks.

Sign In or Register to comment.