Daz Need some opinions and help from experienced devs AND everyone in general that plays vns or know what they are

[Thiefy0]

as the title says , i would love some opinions on the renders and some advice regarding g9 characters .

for almost a year i ve been experimenting and improving my skills as an visual novel dev and i m looking forward to starting developing my first game , but i m in an standstill cause one of my characters a G9F takes a toll on my pc during rendering and it just takes too much time , but i want that character , she looks good , how can i convert her to g8 but still have her look exactly the same and secondhand , what s your opinion on the renders and where i should improve .

PS the game focuses on NTR , Footfetish (a lot) , Socks , Cheating , Coruption and many more genres that i ve seen not get the attention they need , so about a year ago i decided to start a game of my own and here i am a year later writing this , what do you guys think are some top tier fetishes you would wanna see in a game




I KNOW THE RENDERS HAVE GRAIN ON THEM I JUST DIDN T LEAVE THEM FULLY TO DEVELOP CAUSE THEY ARE TEST RENDERS SO DON'T MIND THAT

 

[Turning Tricks]

I'm going to suggest that the problem isn't the generation of the model, but the scene, environment and - most likely - the lighting.

I have used G8's, G8.1's and G9's for many thousands of renders now and for over a year, I only had a 4GB 1050 Ti card, which meant 99% of my renders back them were CPU. And I have never noticed much difference at all in the render time of the actual G9 models.

What I HAVE noticed is that the clothing and newer environments are going nuts with 8K textures and your choices of lighting will heavily influence how fast you can render. As well, he subD level of your characters and objects has a big impact. For those samples above, I would not go above SubD 3. For close-ups, use SubD 4 (if the asset supports that)

Do an experiment... just in your two examples, it seems like you have a few light sources. Most likely a Ghost Light or a Mesh light giving the bright over-all room lighting. The light on the table lamp (emissive or a GL?) and possibly a spot of two (I can actually see two spots reflection on the right example - you need to turn those render emitters OFF).

Take that same scene and remove the Ghost lights and just use 3 spots to light it. Put one spot up high for room lighting. Make it 75 degress wide, Disc, 75cm x 75cm and ... umm 4500K at about 150,000 lumens. Then use two spots to light the model... placed about 90 degrees apart. (one left and right of the camera)

Run that render and compare it to the time for the one with Ghost lights. It's been my experience that when you use large Ghost Lights in a full scene with lots of textures, it really works your GPU.

For example, on your samples, back when I had my shitty 1050 card, I would have rendered the bedroom environment at night on the queue because it would take about 2 hours+ for a decent iteration count. Then I would have rendered the model in real time (take about 20 mins) and put them together in Photoshop. Now that I have a decent GPU, I could do that whole scene in about 12 minutes, give or take. (8000 iterations)

People like to shit on G9's and there's a lot of miss-information that they are so much more resource intensive, but my experience has not backed that up. One of my VN's is 100% exclusively G9's and the other, older one is a mix of G8 and G8.1 and I have seen almost no noticeable time difference in rendering. It's only the newer clothes and environment textures that are getting crazy out of hand, that are impacting render times. Just for example, most of my current renders I am doing are weighing in at 8-12 GB in VRAM, and I am mostly working with G8's on this project.

 

[Thiefy0]

I'm going to suggest that the problem isn't the generation of the model, but the scene, environment and - most likely - the lighting.

I have used G8's, G8.1's and G9's for many thousands of renders now and for over a year, I only had a 4GB 1050 Ti card, which meant 99% of my renders back them were CPU. And I have never noticed much difference at all in the render time of the actual G9 models.

What I HAVE noticed is that the clothing and newer environments are going nuts with 8K textures and your choices of lighting will heavily influence how fast you can render. As well, he subD level of your characters and objects has a big impact. For those samples above, I would not go above SubD 3. For close-ups, use SubD 4 (if the asset supports that)

Do an experiment... just in your two examples, it seems like you have a few light sources. Most likely a Ghost Light or a Mesh light giving the bright over-all room lighting. The light on the table lamp (emissive or a GL?) and possibly a spot of two (I can actually see two spots reflection on the right example - you need to turn those render emitters OFF).

Take that same scene and remove the Ghost lights and just use 3 spots to light it. Put one spot up high for room lighting. Make it 75 degress wide, Disc, 75cm x 75cm and ... umm 4500K at about 150,000 lumens. Then use two spots to light the model... placed about 90 degrees apart. (one left and right of the camera)

Run that render and compare it to the time for the one with Ghost lights. It's been my experience that when you use large Ghost Lights in a full scene with lots of textures, it really works your GPU.

For example, on your samples, back when I had my shitty 1050 card, I would have rendered the bedroom environment at night on the queue because it would take about 2 hours+ for a decent iteration count. Then I would have rendered the model in real time (take about 20 mins) and put them together in Photoshop. Now that I have a decent GPU, I could do that whole scene in about 12 minutes, give or take. (8000 iterations)

People like to shit on G9's and there's a lot of miss-information that they are so much more resource intensive, but my experience has not backed that up. One of my VN's is 100% exclusively G9's and the other, older one is a mix of G8 and G8.1 and I have seen almost no noticeable time difference in rendering. It's only the newer clothes and environment textures that are getting crazy out of hand, that are impacting render times. Just for example, most of my current renders I am doing are weighing in at 8-12 GB in VRAM, and I am mostly working with G8's on this project.

thanks man i really apreciate it , i didn t bother making them look very good but i ll remake the second shot and post it

 

[Thiefy0]

I'm going to suggest that the problem isn't the generation of the model, but the scene, environment and - most likely - the lighting.

I have used G8's, G8.1's and G9's for many thousands of renders now and for over a year, I only had a 4GB 1050 Ti card, which meant 99% of my renders back them were CPU. And I have never noticed much difference at all in the render time of the actual G9 models.

What I HAVE noticed is that the clothing and newer environments are going nuts with 8K textures and your choices of lighting will heavily influence how fast you can render. As well, he subD level of your characters and objects has a big impact. For those samples above, I would not go above SubD 3. For close-ups, use SubD 4 (if the asset supports that)

Do an experiment... just in your two examples, it seems like you have a few light sources. Most likely a Ghost Light or a Mesh light giving the bright over-all room lighting. The light on the table lamp (emissive or a GL?) and possibly a spot of two (I can actually see two spots reflection on the right example - you need to turn those render emitters OFF).

Take that same scene and remove the Ghost lights and just use 3 spots to light it. Put one spot up high for room lighting. Make it 75 degress wide, Disc, 75cm x 75cm and ... umm 4500K at about 150,000 lumens. Then use two spots to light the model... placed about 90 degrees apart. (one left and right of the camera)

Run that render and compare it to the time for the one with Ghost lights. It's been my experience that when you use large Ghost Lights in a full scene with lots of textures, it really works your GPU.

For example, on your samples, back when I had my shitty 1050 card, I would have rendered the bedroom environment at night on the queue because it would take about 2 hours+ for a decent iteration count. Then I would have rendered the model in real time (take about 20 mins) and put them together in Photoshop. Now that I have a decent GPU, I could do that whole scene in about 12 minutes, give or take. (8000 iterations)

People like to shit on G9's and there's a lot of miss-information that they are so much more resource intensive, but my experience has not backed that up. One of my VN's is 100% exclusively G9's and the other, older one is a mix of G8 and G8.1 and I have seen almost no noticeable time difference in rendering. It's only the newer clothes and environment textures that are getting crazy out of hand, that are impacting render times. Just for example, most of my current renders I am doing are weighing in at 8-12 GB in VRAM, and I am mostly working with G8's on this project.

here , what do you think about this one

 

[Turning Tricks]

View attachment 4131737

here , what do you think about this one

That's pretty good.

A couple of suggests to make it even better (IMO)...

Try and get a catch light in her eyes. You can do this by making a Spot that is only about 50K lumens and maybe 10-15 cm size (Disc). Put the spot about 30 degrees to the right of your camera on the same plane. Do an Iray preview and move the spot if the catch light is not quite there. Then you can either do the whole render or just do a spot edit of the areas around the eyes and layer that on the original image (I use the gradient tool to feather in the edges of the spot render ). Conversely, you can just photoshop catch lights, if you know how to do that.

Second, try and adjust the hair to show gravity better. Hair assets are a pain in the ass, since they all have different morphs or are Dforce only, etc. If it's Dforce then run the simulation. If it's dialed-in morphs, try and find one that moves the bangs down to look more natural. My favorite hairs are ones with lots of bones, so I can pick them and adjust them individually (using the Universal Tool). Also, another trick if you have some bangs or something looking totally wacky for the pose you have... if they don't have morphs you can do, you can hide them by selecting them with the Surface Tool and then adjusting the Cutout Opacity to 0. Sometimes to get it perfect, you may have to do two renders... one with the hair showing and the other with it hidden (spot render) and then do some funkery in Photoshop to mesh them together. Learning image editor skills is very important for making this stuff, IMO - especially doing Alpha layer painting.

Finally, you can try using Depth of Field. That's more resource intensive on weaker GPU's though. A trick I used when I had my crappy card, was I would render the normally, then I would copy the background area using the Polygonal Lasso Tool and apply a Gaussian blur to that with a gradient. Basically photoshoping a fake Depth of Field effect.

 

[Thiefy0]

That's pretty good.

A couple of suggests to make it even better (IMO)...

Try and get a catch light in her eyes. You can do this by making a Spot that is only about 50K lumens and maybe 10-15 cm size (Disc). Put the spot about 30 degrees to the right of your camera on the same plane. Do an Iray preview and move the spot if the catch light is not quite there. Then you can either do the whole render or just do a spot edit of the areas around the eyes and layer that on the original image (I use the gradient tool to feather in the edges of the spot render ). Conversely, you can just photoshop catch lights, if you know how to do that.

Second, try and adjust the hair to show gravity better. Hair assets are a pain in the ass, since they all have different morphs or are Dforce only, etc. If it's Dforce then run the simulation. If it's dialed-in morphs, try and find one that moves the bangs down to look more natural. My favorite hairs are ones with lots of bones, so I can pick them and adjust them individually (using the Universal Tool). Also, another trick if you have some bangs or something looking totally wacky for the pose you have... if they don't have morphs you can do, you can hide them by selecting them with the Surface Tool and then adjusting the Cutout Opacity to 0. Sometimes to get it perfect, you may have to do two renders... one with the hair showing and the other with it hidden (spot render) and then do some funkery in Photoshop to mesh them together. Learning image editor skills is very important for making this stuff, IMO - especially doing Alpha layer painting.

Finally, you can try using Depth of Field. That's more resource intensive on weaker GPU's though. A trick I used when I had my crappy card, was I would render the normally, then I would copy the background area using the Polygonal Lasso Tool and apply a Gaussian blur to that with a gradient. Basically photoshoping a fake Depth of Field effect.

man thanks again , you don t even imagine how much you helped me , i apreciate you and wish you the best , sending love

 

[MissFortune]

Try and get a catch light in her eyes

I think getting a good catchlight at that angle may be hard to do. Especially for someone who has little to no experience in lighting (Daz catchlights/eye reflections also tend to not be very accurate). It'd almost certainly be far faster to just add their own in Photoshop:

 

[Turning Tricks]

I think getting a good catchlight at that angle may be hard to do. Especially for someone who has little to no experience in lighting (Daz catchlights/eye reflections also tend to not be very accurate). It'd almost certainly be far faster to just add their own in Photoshop:

View attachment 4133834

Ironically, I just was having some difficulty getting a good catch light on a scene yesterday. I got it, but only after messing around with a spot light for 20 mins.

To me, it's like playing pool. You have to figure out the right angle for the spot to "bounce" off the spherical eyeball and hit the camera lens. The other factor is the size and shape of a spot. I find 10-15cm rectangle or disc works best. Normally my spots are 50-100 for a nicer shadow, but they don't work good as catch lights.

I mentioned the two ways because some people are hesitant to play with lights much in DAZ while others are terrified of Photoshop. PS has some nice templates and brushes for catch lights though.