Somethings to Consider when starting to learn CG

24567

Comments

  • SzarkSzark Posts: 10,634
    edited December 1969

    Thanks tunggul1101.

    Does a happy dance with the feedback I am getting, makes the hours spent typing it up worth while. Yes hours and hours because I am Dyslexic and writing is my therapy.

  • adaceyadacey Posts: 186
    edited December 1969

    Here's my very basic tips for how I work with tweaking things.

    1. I usually work with each figure or major part of the scene as their own scene file first. I'll add the clothing and props for the figure and then start tweaking the surfaces. Doing most of this "heavy lifting" with very basic scenes allows me to work a lot more quickly for fine-tuning without having to render the whole scene, even when just using the spot render tool. Once I'm happy with the individual components I'll merge them together into a final merged scene where I'll take care of the posing and positioning.

    2. While working on surfaces add at least 1 light with shadows turned on (I prefer ray-traced shadows). Usually I'll just use 1 distant light positioned at an angle to the figure so that the light glances across the surfaces. You need to have at least 1 light off the axis of your viewpoint (be that a camera or perspective view) in order to see effects of your changes to things like bump, displacement, specular, etc. The headlamp is useless for showing these effects since it's always directly on axis with the viewpoint so you won't see shadows or specular very well.

    3. Only adjust 1 thing at a time! Make 1 change and do at least a spot render on the surface you're working on. If you change multiple things at once you'll never be able to make sense of what each setting is doing. So try something like apply a map, spot render, adjust the strength slider for that same setting, spot render, dial in your settings for that 1 section before you move along. That's not to say that as you add other settings that you won't have to go back and tweak things but at least you'll have a much better sense of what each section is contributing to the final scene and what modifying the sliders or maps will do.

    4. Don't be afraid to edit the maps that came with your textures to create the maps you need. Convert a diffuse map to grayscale, play with contrast, invert the image. Even just those basic edits can add a lot to the final look without you having to start from scratch.

    I still feel like I'm barely scratching the surface with my editing but I thought I'd share what I've learned so far in case it's helpful to others.

  • SzarkSzark Posts: 10,634
    edited December 1969

    First sorry for not responding before now I wanted to give the respect of answering properly which takes time and a clear head. Now that I have had some sleep but full of a head cold I can reply. :)

    adacey said:
    Here's my very basic tips for how I work with tweaking things.

    1. I usually work with each figure or major part of the scene as their own scene file first. I'll add the clothing and props for the figure and then start tweaking the surfaces. Doing most of this "heavy lifting" with very basic scenes allows me to work a lot more quickly for fine-tuning without having to render the whole scene, even when just using the spot render tool. Once I'm happy with the individual components I'll merge them together into a final merged scene where I'll take care of the posing and positioning.
    This is exactly how I work now but in the beginning I had to have everything laid out in front of me. This was due to not having a vivid imagination given I was starved of any creative outlet for over 30 years. The consequence of this picturing an image in my mind was just not working right.
    Now that I can pictures things I do what you do, split it up in to Elements but given the similar workflow we differ on the lighting.

    adacey said:
    2. While working on surfaces add at least 1 light with shadows turned on (I prefer ray-traced shadows). Usually I'll just use 1 distant light positioned at an angle to the figure so that the light glances across the surfaces. You need to have at least 1 light off the axis of your viewpoint (be that a camera or perspective view) in order to see effects of your changes to things like bump, displacement, specular, etc. The headlamp is useless for showing these effects since it's always directly on axis with the viewpoint so you won't see shadows or specular very well.
    The first thing I do is set the scene first, background or building etc, load and position the props, then I set up all the lighting. Then with the new Save As > Scene Subset in DS4.5 I save the lights as a scene or depending on the lights, the old Lighting Preset. I then use that light set to set up my figures. But I must admit doing this way you do have to have an idea on where your main elements of the image are. But it works for me.

    Once I have the lights reasonably setup I get stuck in working on the surfaces using your advice below. Yes it takes a long time but that is what you have to do to get better quality, if that is what you want. You may opt for postwork like I am doing these days. Rendering layers and merging them in PS and because I am not bothering about realism I have the freedom to try all sorts of things without being bound by reality.


    adacey said:
    3. Only adjust 1 thing at a time! Make 1 change and do at least a spot render on the surface you're working on. If you change multiple things at once you'll never be able to make sense of what each setting is doing. So try something like apply a map, spot render, adjust the strength slider for that same setting, spot render, dial in your settings for that 1 section before you move along. That's not to say that as you add other settings that you won't have to go back and tweak things but at least you'll have a much better sense of what each section is contributing to the final scene and what modifying the sliders or maps will do.
    I think for many new users, me included, see these stunning images made by others and want to be off and running straight away. Well computers supposed to make things easy...yeah right. :) I found it better to learn about the basic Surfaces first, to the point I knew exactly what each channel did, how it worked and why. I did this before I started to delve in Advanced shaders like pwSurfaces etc and Uber Surface etc. Take this image called “Neglect” were I said

    The first image was made back in April 2011 coming first for another Dreamlight’s monthly contest themed “A Dogs Life”. Gave myself a challenge with this one. I wanted to see what I could do with only the maps that came with the props. I usually add some additional maps to surfaces when I do an image. I then plugged in Bump, displacement and diffuse maps in all sorts of channels testing the results. If it didn’t have displacement I would use the bump if it had neither I would use the diffuse map.
    What I didn’t say is that it took me 3 weeks of testing settings of all the surface channels, change one setting, test render, change it back, change another, test render again and so on for weeks. OK the rain effects are postwork but all the surfaces got the treatment. Here is what it looked like before and after, without postwork. Both images below have the same lights at the same settings but with the default texture settings and maps in one and the other well a big difference with just playing with surfaces.

    4. Don't be afraid to edit the maps that came with your textures to create the maps you need. Convert a diffuse map to grayscale, play with contrast, invert the image. Even just those basic edits can add a lot to the final look without you having to start from scratch.


    Because I had been learning PSE for a number of years prior to Daz Studio my very first image had me looking at the texture maps. I firmly believe knowing image manipulation software like Photoshop CS/Elements and Gimp really helped me to learn about texture maps as I wasn’t scared to delve in to the surfaces pane from the get go.

    I still feel like I'm barely scratching the surface with my editing but I thought I'd share what I've learned so far in case it's helpful to others.I am glad you did you gave some good tips, thank you.

    Without_postwork.jpg
    1000 x 562 - 294K
    NeglectPlain.jpg
    1010 x 568 - 382K
  • adaceyadacey Posts: 186
    edited November 2012

    Szark said:
    First sorry for not responding before now I wanted to give the respect of answering properly which takes time and a clear head. Now that I have had some sleep but full of a head cold I can reply. :)

    No worries, thanks for taking the time to do such a detailed response.


    This is exactly how I work now but in the beginning I had to have everything laid out in front of me. This was due to not having a vivid imagination given I was starved of any creative outlet for over 30 years. The consequence of this picturing an image in my mind was just not working right.

    Fair enough, there's pros and cons to each approach and I suppose a lot depends on how your mind works. The main reason I mentioned this workflow in the context of this thread was because we were discussing learning how to get better images and I find that isolating your elements really helps with tweaking surfaces both because it removes distracting elements and because your test renders will go more quickly.


    The first thing I do is set the scene first, background or building etc, load and position the props, then I set up all the lighting. Then with the new Save As > Scene Subset in DS4.5 I save the lights as a scene or depending on the lights, the old Lighting Preset. I then use that light set to set up my figures. But I must admit doing this way you do have to have an idea on where your main elements of the image are. But it works for me.

    That's a really good tip and I hadn't realized there was the option to save a subset of the scene. For me, I have a lot of experience in photography so maybe I'm just better able to visualize how different lighting will impact the surfaces (although, I'm still struggling with how biased rendering's lighting differs from real life). Or, the other factor could be that I try to make every surface look as good as I can get it, so I'll often move the lighting around to highlight the particular surface. I think part of this is due to me working on each element individually first, sometimes I don't know what's going to be visible in the scene or not so I'm trying to get everything looking as good as I can get it. I think the other aspect is that I'm coming from a photo background and haven't quite come around to the idea that I can turn off things that won't impact the final scene.


    Once I have the lights reasonably setup I get stuck in working on the surfaces using your advice below. Yes it takes a long time but that is what you have to do to get better quality, if that is what you want. You may opt for postwork like I am doing these days. Rendering layers and merging them in PS and because I am not bothering about realism I have the freedom to try all sorts of things without being bound by reality.

    I've played a bit with rendering each light separately, using a workflow technique that was shown on the Dreamlight youtube channel. Being pretty familiar with Photoshop I found that it was really easy to change the intensity and colour of the lighting in post and not worry at all about getting it right in the render. I haven't played with layering further than that so I'd love to hear more about what you do with your different render passes.


    Because I had been learning PSE for a number of years prior to Daz Studio my very first image had me looking at the texture maps. I firmly believe knowing image manipulation software like Photoshop CS/Elements and Gimp really helped me to learn about texture maps as I wasn’t scared to delve in to the surfaces pane from the get go.

    Agreed, there's so much power there if you're comfortable with working in an image editor, both for tweaking textures and for post work, I'd almost call it an essential skill if you want to get the best out of your renders.

    Post edited by adacey on
  • SzarkSzark Posts: 10,634
    edited December 1969

    adacey said:
    No worries, thanks for taking the time to do such a detailed response.
    Well this thread deserves more than simple answers and also since you took the time to help, well it is just the right thing to do. :)
    adacey said:
    Fair enough, there's pros and cons to each approach and I suppose a lot depends on how your mind works. The main reason I mentioned this workflow in the context of this thread was because we were discussing learning how to get better images and I find that isolating your elements really helps with tweaking surfaces both because it removes distracting elements and because your test renders will go more quickly. Oh yeah I do agree entirely, sorry I get side tracked. Also it can free up memory when like me have little of it. No not my memory the computer’s. ;)

    adacey said:
    That's a really good tip and I hadn't realized there was the option to save a subset of the scene. For me, I have a lot of experience in photography so maybe I'm just better able to visualize how different lighting will impact the surfaces (although, I'm still struggling with how biased rendering's lighting differs from real life). Or, the other factor could be that I try to make every surface look as good as I can get it, so I'll often move the lighting around to highlight the particular surface. I think part of this is due to me working on each element individually first, sometimes I don't know what's going to be visible in the scene or not so I'm trying to get everything looking as good as I can get it. I think the other aspect is that I'm coming from a photo background and haven't quite come around to the idea that I can turn off things that won't impact the final scene.

    Interesting you should say this about your photographic background as I often here that many Photographers find learning CG lighting easier than some non Photographers. Anyway biased render engine has to fake everything based on the surfaces and how they react to light, ray tracing. So people have thought of clever ways to fake Radiosity by understanding Renderman compliant shaders like Uber Environment 2. :) Or Making shader indirect lights and cameras using the Shader Mixer which I have done but my computer can’t handle the amount of calculations needed to render a scene at high quality setting which you need for advanced lighting. Or using spot lights for Key, fill, bounce and even specular highlights only.

    I've played a bit with rendering each light separately, using a workflow technique that was shown on the Dreamlight youtube channel. Being pretty familiar with Photoshop I found that it was really easy to change the intensity and colour of the lighting in post and not worry at all about getting it right in the render. I haven't played with layering further than that so I'd love to hear more about what you do with your different render passes.


    Multi-pass rendering where you render an ambient layer, ambient occlusion layer and Key light layer etc is pretty much what Dreamlight shows but different to what I do. This image http://www.daz3d.com/forums/discussion/2765/P120/#166060 was made with three renders this is due to not enough memory. The first main image is the room and props, the second the poor bloke laid out and the room and the third, the female figure and room, all with the exact same lighting. I then made a B&W mask in Daz Studio using Surface Mask Creator of the women and the male figure and what he is laid on.
    4 layers: On the bottom the room and prop render (there is reason for this). Next and male figure and prop and then two copies of the female figure render. I then applied the masks to the male layer and one of the female layers. For the other woman layer I erased everything except her shadow. So we have her nicely cut out and her shadow underneath. This shadow layer was merged down to the base layer which would have affected how the next layer I made worked if I hadn’t merged it down. Yes you do have to do some thinking on where shadows fall and plan accordingly.

    So I ended up with 3 layers, the base with her shadow now visible as part of the background layer, a masked layer of the male and what he is laid on and a masked layer of the female including her clothing, hair and weapon of choice.

    Ok why the base layer...I wanted to be able to control how the next layer works. I made an Ambient Glow render next. I deleted all the lights barring one spot but turned it off. I found if you don’t keep at least one light in the scene the preview light gets rendered even with turning it off. Anyways the viewport should be pitch black but with any surfaces that has Ambient applied in the surfaces pane will show up, glow in effect. See image below.
    I placed this layer on top of the base layer, below the male and female layers. Set it to Screen, copied and applied 8% Gaussian blur to the copy, duplicated this layer and applied 15% Gaussian Blur, duplicate > 25% Gaussian Blur. This gives that fuzzy light look. If I didn’t make the masks then this effect would spill over the figures too much.

    Agreed, there's so much power there if you're comfortable with working in an image editor, both for tweaking textures and for post work, I'd almost call it an essential skill if you want to get the best out of your renders.My thoughts exactly for which I think may would agree on.

    ambient.jpg
    1400 x 787 - 72K
  • adaceyadacey Posts: 186
    edited December 1969

    Szark said:

    Interesting you should say this about your photographic background as I often here that many Photographers find learning CG lighting easier than some non Photographers. Anyway biased render engine has to fake everything based on the surfaces and how they react to light, ray tracing. So people have thought of clever ways to fake Radiosity by understanding Renderman compliant shaders like Uber Environment 2. :) Or Making shader indirect lights and cameras using the Shader Mixer which I have done but my computer can’t handle the amount of calculations needed to render a scene at high quality setting which you need for advanced lighting. Or using spot lights for Key, fill, bounce and even specular highlights only.

    Struggle might be a little too strong a word. The challenge I find is that I think I have a pretty decent understanding of how light works in the real world and sometimes it can get frustrating working with a biased render when you know how things would work in real life but can't get it to work that way in your render. The big ones I can think of off the top of my head would be light falloff (fixed with using the right lights, like the various uber lights that have adjustable falloff) and bounce, which as you mentioned has solutions. Similarly, there's some differences in how light behaves. For instance, in real life, the size of the light and distance from the subject will determine how soft or hard the light is, in a biased renderer you can adjust the shadow softness but it's a global change. Of course the solution to all of this is to use Reality but that has its own challenges with getting surfaces to look good and can result in much higher render times.

    Another good example was that recently I was working on a render for a challenge on this forum. I didn't want to use Reality because of how long it would take to render the scene and because I'd done a fair amount of work adjusting the surfaces in DAZ's shaders and knew that wouldn't translate over. What drove me nuts was the scene was a nighttime indoor scene where I wanted to have "moonlight" shining through a window as the main light. The moonlight was easy (distant light placed outside of the window at the angle I wanted, could have used a spotlight for that matter but there was a ceiling on this prop). What was incredibly challenging for me was getting some type of ambient light level into the room, I tried UE in global illumination mode and it was almost pitch black (not enough light from the window). I ended up having to really dial in the settings in Uber SoftLight to get the look I wanted and overall I'm very happy with it but it was a big challenge to get that. What was frustrating for me was that I knew exactly how I'd light the scene in Reality but those techniques wouldn't work in a biased renderer.

    Multi-pass rendering where you render an ambient layer, ambient occlusion layer and Key light layer etc is pretty much what Dreamlight shows but different to what I do.

    The method I meant was this one. The video talks about using it for saving rendering time. What I realized was that if I was going to be layering the lights then there was no reason to adjust the intensity of the light or the colour, you can just render each light as white and at 100% intensity. Or, you only need to adjust the intensity if you're blowing out your highlights.

    You still layer each light the same way as described in the video, but then you can apply adjustments to each light individually to get the colour you want, no need to re-render the scene. I'll post an example when I'm on my main system to show what I'm talking about.

  • adaceyadacey Posts: 186
    edited December 1969

    Okay, at my main computer now. Here's an example of the layering I did. There were 4 spotlights in this scene I rendered each one individually and had the intensity set to 100% for each and the colour set to white.

    I converted each image into a smart object (not essential but it makes your adjust of each layer a little easier without the need to keep masking every adjustment). In that smart object, I applied the adjustments for that layer, keeping the background black. For this image, I used the photo filter adjustment layers to adjust the colour temperature of each light. I then layered them all together and set the blend modes to screen for each, adjusting the transparency of each smart object to adjust that light's intensity. I then applied my post work on that final layered file (probably merged it all into another smart object).

    Troll-Fight.jpg
    1200 x 900 - 478K
    Bounce.jpg
    1200 x 900 - 211K
    Rim1.jpg
    1200 x 900 - 99K
    Fill.jpg
    1200 x 900 - 89K
    Main.jpg
    1200 x 900 - 213K
  • SzarkSzark Posts: 10,634
    edited November 2012

    To get ambient light in to a room I use UE2 set to Ambient Only or Occlusiion with Soft shadows or use 8 distant lights set to about 1 - 1.5 % intensity no Shadows, 4 pointing downward facing NSE and West and the other 4 pointing updward, 45% angle for all 8.

    There so many wasy to do the same thing it does make it hard to choose one method. Plus it depends on the scene. Nightime UE2 Ambient only, 10% Intenstity, bluish medium statruation light clour, work well for a nice full moon ambient light. Distant with a light blue/gry colour at about 15% Shadow mapped shadows.

    I have used the twice so far in my last two images...haven't posted the second yet.

    Yes I have seen the DL vid and again there are so many ways. I have seen this layering before as you show which is very cool but it does require having a good Photoshop/GIMP etc knowledge and unerstanding. I do like the end result as that is the point...the end results how we get there doens't really matter.

    Thanks for sharing and contributing to this thread adacey... It is nice that I am also getting to learn new things.

    Post edited by Szark on
  • adaceyadacey Posts: 186
    edited December 1969

    Thanks for the tips on the indoor ambient. I was thinking that something like this would be the solution but I'd only used 1 distant light which would explain why it wasn't working that well.

    The layering images I showed were actually from one of my first finished image in DAZ (if not the first, definitely the first one I'd shared publicly). I'd played around with it for awhile before then and I'd also dabbled in 3-d way back in the 90's so it wasn't completely new to me.

    In that case, I had dialed in the light placement but was playing quite a bit with getting the balance and the colour right. I'd stumbled across the Dreamlight video about rendering each light separately and layering it and I realized this was the solution to my problem. For me, getting the colour and intensity right in Photoshop was pretty easy.

    I think what we're really getting at is there are a lot of different solutions to the same product, and as you said, it's the final product that really matters. How you get there is really a question of where you put your effort and what can get you there the best way you can. If you can get the effect quickly in the render, great. If it's faster for you to fix something in post, go for it.

  • 2Fatbear2Fatbear Posts: 49
    edited December 1969

    adacey said:
    I converted each image into a smart object (not essential but it makes your adjust of each layer a little easier without the need to keep masking every adjustment). In that smart object, I applied the adjustments for that layer, keeping the background black. For this image, I used the photo filter adjustment layers to adjust the colour temperature of each light. I then layered them all together and set the blend modes to screen for each, adjusting the transparency of each smart object to adjust that light's intensity. I then applied my post work on that final layered file (probably merged it all into another smart object).

    Hi Adacey

    What is a "smart object"? Or, can you point me to where I can find more info on this technique youre using here? I know a bit about Photoshop/Gimp so no problem on this area, I just dont know about smart objects and exactly what are "photo filter layers". Is this plain editing on Photoshop?


    Its really great to learn how you two are building your images, the references to how you use lights and etc, a great lesson to a newbie like me.

  • adaceyadacey Posts: 186
    edited December 1969

    2Fatbear said:

    Hi Adacey

    What is a "smart object"? Or, can you point me to where I can find more info on this technique youre using here? I know a bit about Photoshop/Gimp so no problem on this area, I just dont know about smart objects and exactly what are "photo filter layers". Is this plain editing on Photoshop?


    Its really great to learn how you two are building your images, the references to how you use lights and etc, a great lesson to a newbie like me.

    Smart objects are a feature that got added in Photoshop a few versions ago (CS3 or 4 I think, I could be wrong as I went from CS on an old computer to CS5, and now 6, on my current machine). Quick tutorial is that you can select 1 or more layers in PS and then right click on them and select "convert to smart object". This will make it look like they're all merged into a single layer in your document but you'll see an icon next to the layer's preview that shows it's a smart object. If you double click on the preview it will open up that smart object in a separate window (looks like you're working on a separate PS document). Any changes you make there will show up in your main document when you save this. So it's almost like you're embedding a separate PS document into your "main" PS document.

    Now, why would you do all of this? Well the huge feature of smart objects is they allow you to use smart filters. Most (if not all) of PS's built-in filters (and several 3rd party filters) will work with smart objects. What smart filters do is save the filter settings you applied to the smart object, and they'll show up listed below the layer. At any time, you can double click on that filter to get back into the filter and change your settings, when you hit okay it will update the layer (may take some time if it's a big document or has several smart filters applied). This is a really powerful workflow and I recommend giving it a try.

    Now the reason I did it was that I wanted to work with each light separately with adjustment layers. To answer your 2nd question, "Photo Filter" is an adjustment option that can be applied in the Image -> Adjustments -> Photo Filter, or apply it as an adjustment layer (my preference). It offers you the effect of several traditional photographic adjustment filters that you'd put over your lens while shooting, most of them relating to adjusting the colour temperature of the light from the film days when you couldn't just change the white balance on your camera. I used the warming and cooling filters from this selection to warm up most of the lights (wanted it to look like torch light) while cooling the light that was coming from outside (mimicking night light).

    The reason I used smart objects in this case was because I didn't want to have to mask each adjustment layer to control how they impacted the light. If I'd just laid the adjustment layer over all of my lights then it would impact everything that was visible. So, I made each light's layer a smart layer and put the adjustment layer in each of those objects, that guaranteed that it only impacted that light. I then set each of the smart object layers to the screen blending mode to mix them together and adjusted the opacity to adjust the intensity of each light. Not saying this is the only way to do that, but it's the way that worked for me.

    In the end, I selected all of these smart objects and converted them into yet another smart object which I applied several filters to, taking advantage of the smart filter functionality to keep the flexibility intact, to get the final look.

    Hope this helps, let me know if you need more info as I know I've sort of gone through this quickly.

  • adaceyadacey Posts: 186
    edited December 1969

    Szark, just wanted to comment quickly on the differences in our approaches (pose and light first, then save subsets vs. start with each element separately then merge). On my last scene I spent ages working on individual elements first, then merged these elements for arranging and posing in the final scene. What I found after doing this was that I'd spent a ton of time working on tweaking surfaces that were either not visible in the final render or were far enough away from the camera that the fine details really didn't show up. This meant that a best, all of that work was wasted time (not visible) and at worst, just meant an increased render time for little to no benefit.

    What I realized was that I'm still thinking of trying to make each piece look as good as it possibly can and am not necessarily thinking of the final image enough. Your approach, where the positioning and lighting is done first, would help quite a bit for this if you're working on a single image. My approach seems like it might be better suited for either animation (where you're going to see the element from more than 1 angle) or if you were building up a library of objects/characters/whatever that you were intended to continue to reuse over a bigger project and wanted them to look consistent (say if you were rendering a comic or something).

    Not that the time was completely wasted, I still learned a ton and did improve the overall look of the image, but I did come down to the wire on finishing the image for the RRR contest both because I'd spent so long tweaking surfaces and because the final render took ages to complete because of all that detail I'd added.

  • SzarkSzark Posts: 10,634
    edited December 1969

    Sorry I haven't replied earlier, I am so busy I can't find much time for myself.

    adacey...Nice info on Smart Objects, don't have that in PSE6...I think it is included in PSE11 now

    As for setting up surfaces that aren't seen...I hear you on that one. Now I am always thinking about removing textures I can't see in the viewport. Removing bump and displacement maps for distant items...all in the effort of saving memory etc etc.

    2Fatbear nice that you are getting something from this thread.

  • 2Fatbear2Fatbear Posts: 49
    edited December 1969

    Adacey, many thanks for the tutorial, I surely will use it a lot.

    Its interesting that this approach of yours with smart objects is similar to what I do with Reality/Lux, to work lights intensity and color while rendering when I can see all shadows properly. Lux also has a list of kinds of old films aspects we can apply to the image, similar to what you are doing with photo filter (but I'm guessing photo filter will have more resources than this).

    I started playing with Uber-Things thinking in cutting the huge time Lux takes to render a big image. But then I noticed that when we use UE2 on its full power the time will be also huge (still to do a proper comparation), so maybe your idea will help lower render time since youre rendering only a light for each layer (or are you? Or there are the eight distant lights on each layer?)

    Anyway, I know now we just cant do great renders in one minute (or sixty! (: ), which makes even more valuable the topic of this thread, we have to work very well all details before going to the final render.

  • SzarkSzark Posts: 10,634
    edited December 1969

    2Fatbear said:

    Anyway, I know now we just cant do great renders in one minute (or sixty! (: ), which makes even more valuable the topic of this thread, we have to work very well all details before going to the final render.
    Exactly, this is why it takes me weeks if not months to do a realistic image. :)
  • adaceyadacey Posts: 186
    edited December 1969

    2Fatbear said:

    Its interesting that this approach of yours with smart objects is similar to what I do with Reality/Lux, to work lights intensity and color while rendering when I can see all shadows properly. Lux also has a list of kinds of old films aspects we can apply to the image, similar to what you are doing with photo filter (but I'm guessing photo filter will have more resources than this).

    I started playing with Uber-Things thinking in cutting the huge time Lux takes to render a big image. But then I noticed that when we use UE2 on its full power the time will be also huge (still to do a proper comparation), so maybe your idea will help lower render time since youre rendering only a light for each layer (or are you? Or there are the eight distant lights on each layer?)

    Anyway, I know now we just cant do great renders in one minute (or sixty! (: ), which makes even more valuable the topic of this thread, we have to work very well all details before going to the final render.

    Yeah, I love Reality but I've been pushing myself to learn more native DAZ lighting and surfaces. Right now, I feel like I have a better handle on lighting in Reality (using my photo background) but a better handle on surfaces in DAZ so it's a bit of a toss up on which way I want to go. I've heard of some tricks for speeding up renders in Reality but I've yet to try them out yet so I can't really comment on whether that will help things out.

    For the example I showed above I did render each light individually, the images shown are the layers for the 4 lights I used (all spotlights); main, fill, bounce and rim. The final image was the merged image (with the colours adjusted) with some additional filters applied. Rendering 1 light at a time can make a big difference in your render times so if you're comfortable with working with the layers in post it can be a great way to get quicker renders and better control. The more lights you're using the more benefit you'll see but then you do need to be pretty organized with your renders so that you don't lose track of which ones you've already done and then there's more layers to work with in PS too.

  • SzarkSzark Posts: 10,634
    edited December 1969

    Yes Organisation and naming conventions realy are important when using layers. Some of my images have had 70 odd layers building up effects etc. This is so I have complete contol over every detail. I so got used to naming all the layers as I make them.

  • adaceyadacey Posts: 186
    edited December 1969

    Szark said:
    Yes Organisation and naming conventions realy are important when using layers. Some of my images have had 70 odd layers building up effects etc. This is so I have complete contol over every detail. I so got used to naming all the layers as I make them.

    That's 1 advantage of smart objects. In the examples I posted I had 1 smart object for each light's layer with its respective adjustment layer as well, so it would then just look like 4 smart objects instead of 8 layers, if I were going even more complicated then it'd be handy. The downside is you then have to open the smart object to access the layers that are inside it. In retrospect, I think I could've gotten away with using layer sets (e.g. folders for the layers) for each light and its respective adjustment layer and then set the blend mode on the layer set, which is another feature that got added at some point (you used to only be able to adjust the opacity of the layer set).

  • SzarkSzark Posts: 10,634
    edited December 1969

    Good if you have the Photoshop CS range unlike many of us here as we have to use less pricey image software like PSE (Photoshop Elements, version 6 in my case) or GIMP etc.

  • adaceyadacey Posts: 186
    edited December 1969

    Szark said:
    Good if you have the Photoshop CS range unlike many of us here as we have to use less pricey image software like PSE (Photoshop Elements, version 6 in my case) or GIMP etc.

    Definitely, I bit the bullet and got the full PS but PSE or the GIMP are definitely viable options. The new subscription options that Adobe's offering takes the sting out a little bit but then you have the recurring charges every month, and you lose the software if you ever stop your subscription so there's pros and cons.

  • SzarkSzark Posts: 10,634
    edited December 1969

    Because I am using PSE and I cannot apply Masks (like 10 and 11 can) so I use GIMP instead and have done so for a while now. But because I know PSE inside and out and tend to apply masks in GIMP save out as a PSD and finish off in PSE. But I am starting to use GIMP more and more now with 2.8 and UI customisation. I know have GIMP looking like my PSE does makes it so much better on the eyes too.

  • Scott LivingstonScott Livingston Posts: 4,331
    edited December 1969

    I just found this and it looks extremely helpful. Thanks so much for putting it together!

  • SzarkSzark Posts: 10,634
    edited December 1969

    Thanks Scott.

    I was planning on doing some more over the course of December but I have to man the December's New Users Contest My plan is to have something done toward the end of Jan next year.

  • 2Fatbear2Fatbear Posts: 49
    edited December 1969

    Szark said:
    Thanks Scott.

    I was planning on doing some more over the course of December but I have to man the December's New Users Contest My plan is to have something done toward the end of Jan next year.

    We wait anxiously (:

    {}fatbear

  • SzarkSzark Posts: 10,634
    edited December 1969

    LOL 2Fatbear not too anxious I hope.

    Here is what has been taking my time recently http://www.daz3d.com/forums/discussion/12654/ some real newbie art info, cut down for obvious reasons but it might help if you are new to art in general.

  • fakedoctorfakedoctor Posts: 30
    edited December 1969

    This thread has been invaluable to me now that I'm really getting into all this stuff, so first of all, thanks for the time you put into it Szark.

    I do have one question, though, reading the link on cameras and lighting. What the heck is "shadow banding?" The author implies that it's a big deal and you don't want it to happen, and says that it is "obvious" in one example picture.. but I can't tell what he's talking about at all. And Google gives me nothing but people asking how to "fix shadow banding" in one program or another, with no indication of what is actually in need of fixing.. I feel like it's something I'm supposed to know, heh.

  • SzarkSzark Posts: 10,634
    edited December 1969

    Another happy customer. My pleasure asd000
    Good question. I know what it is and how it comes about but this is going to be a hard one to get across. Unlike the first page of this thread this is hard for me to put in to words without referring to, what jaderail calls geekish terms. :) Plus I am under the whether a bit today.
    Shadow Banding: Take a look at the second pic in this post http://www.daz3d.com/forums/discussion/12943/P15/#187102 Also banding can occur with Soft Shadows where the shadows go from dark to light in sharp bands of contrasting shades so the term can mean different things but the first example is the most common IMHO. However using Ray Traced Shadows and adjusting the Shadow Bias will fix the issues.

    Deep Shadow Maps and are just that, maps and I think the resolution in DAZ Studio for DSMs is 1024 x 1024. Shadows are pre-calculated before rendering and overlaid on to the main rendered layer. We don't this take place as it is all done by magic elves in our computer. So if your image is a higher resolution than 1024 x 1024 then the shadows will be somewhat poor because of the resolution conflict.

    Ray Traced Shadows are calculated and rendered together with the whole image and are more precise with how they are calculated hence taking longer to render. In simple terms a ray is sent out, it detects a surface, calculates the properties of the surface, and is it reflective, dull etc, and renders the surface according to its settings and calculates where the shadows fall etc. Ray Traced is also connected to Max Ray Trace Depth (MRTD) in the Advanced Render Settings. With Ray Traced Shadows though a MRTD value of 0 will still casts shadows. But if say have an empty glass with thick sides (not just a simple skin mesh) in the scene and you are using Refraction and Ray Traced Shadows then the MRTD should be increased to how many surfaces the Ray has to go through before it passes all the way through the object and casts a shadow or even caustics. In the case of the empty glass it has 4 surfaces for the Ray to pass though, 2 on each side so this gives us our MRTD value, 4. If the glass had liquid in then add 1 to the Max Ray Trace Depth setting making the overall value, 5. This is a long subject and as you can see the Advanced Render Settings are key when dealing with light, shadows and surfaces.

    So when to use DSM or Ray Traced well it goes back to what you want to do. Ray Traced Shadows are more suited to realistic or higher quality images in general. But many around here use DSMs as they generally do a lot of postwork and given the more artistic side of these images accuracy doesn't matter so much.


    So when to use DSM or Ray Traced well it goes back to what you want to do. Ray Traced Shadows are more suited to realistic or higher quality images in general. But many around here use DSMs as they generally do a lot of postwork and given the more artistic side of these images accuacy doesn't matter so much.

  • adaceyadacey Posts: 186
    edited December 1969

    One note about deep shadow maps, I've noticed that in some circumstances they just don't work right. I seem to recall reading there's an issue with point lights and deep shadow maps, not sure if that's been fixed or not but I have seen it in the past. I had a linear point light in a lantern and when it was on deep shadow map the outside of the lantern was being illuminated by the point light. As soon as I switched it to raytraced shadows the problem went away.

    I've seen some other weird issues with certain types of lights too in specific situations, so in some cases don't be afraid to test something and if it seems like it's not working the way you think it should try tweaking the settings or try a different light type and see if the problem persists. Not saying it's a guarantee but I've seen it a few times.

  • SzarkSzark Posts: 10,634
    edited December 1969

    Quiclky before I go to bed point light never have worked with Deep Shadow Mapping and far as I am aware it has always been like that and will continue to.

  • adaceyadacey Posts: 186
    edited December 1969

    Szark said:
    Quiclky before I go to bed point light never have worked with Deep Shadow Mapping and far as I am aware it has always been like that and will continue to.

    Yeah, that's what I found out. That really put me off deep shadow maps at the time and I haven't gone back to them since.

    The other really odd issue I came across was the lights in the UberSoft kit don't seem to work with transparency. I tried putting an ubersoft distant light to shine through a window and it didn't work, which drove me nuts for awhile until I figured that out. Then later, I tried an ubersoft point light in a lantern (seems to be a recurring theme for me) and the inside of the lantern lit up but nothing outside of the lamp was lit. By that time, I'd already determined the transparency issue so I just switched it for an Uber Point light instead, which worked fine.

Sign In or Register to comment.