Floaty Gun Firing Range Short Film

Current Status: City Environment!

I’ll come up with a better title later. I’ve decided to create a dedicated thread for my next short film. I’ve already posted a bit about it in my sketchbook, but I don’t want to flood that thread with images of the same project, so I’ll post about it here instead.

The goals for this project are:

  1. Learn how to use the dirtmap script to quickly create convincing materials for complex models.
  2. Learn how to create believable environments with Agisoft PhotoScan.
  3. Create an easy to use gun rig, with a system for simulating spent shells (up to thousands of them).
  4. Make a cool video!

Above is a little bit of look development I’ve started, using the gun model and a preliminary version of the scanned environment. That environment model is from about 3 weeks ago, and I’ve since shot new images (over 350 of them!) that I’ll be creating new set pieces from, which will have higher detail. You can see some of my other tests with the PhotoScan program here:


All texture work is created using the dirtmap workflow; only dirtmaps and tileable textures are utilized.

Really Cool stuff, inspiring me! What’s this dirtmap script you mention? Sounds like something I could put good use to.

Looks good. I’m impressed.

Looking forward to seeing more, James. Looks good already.

Hey, James, good to see a new thread for this - although I was following your sketchbook :slight_smile: I think in the WIP section of the forum we can discuss the pros and cons more in depth, because your sketchbook is primarily meant to show your work.

Now I ask myself for some time already, how much control you have over the dirtmap script, meaning where it places the dirt. In some areas the dirtmap workflow created convincing results, but here are this big brighter areas (especially at the font grip) which are not self explanatory and which don’t follow the geometry of the gun close enough imho.

But using something “procedural” like this is a promising approach and thanks for sharing that, James!

Thanks for sharing james. Texturing is my lacking skill set right now. Trying to find what tools work best. Been working with ddo2 the last few nights, and I really like it. The one issue I have with it, is how memory intensive it is. Being able to generate some procedural dirtmaps would save me time. I’m off of work the next four days. My sketchbook will have some new entries for sure.

Thanks for sharing james. Texturing is my lacking skill set right now. Trying to find what tools work best. Been working with ddo2 the last few nights, and I really like it. The one issue I have with it, is how memory intensive it is. Being able to generate some procedural dirtmaps would save me time. I’m off of work the next four days. My sketchbook will have some new entries for sure.

I’ve rendered out a small “Look Development” test and added it to my original post. I’ve not done this before, but I believe that by figuring out how things will look early on, I’ll be able to create a better final video.
Straight away, I know now that the gun needs more attention devoted to the fact that it’s basically serving as a character in this video. It could benefit from some sort of hovering movement, maybe a kind of light underneath it to act as a means of support. A laser sight could be added to the sides to serve as “eyes”. These lights should power up to indicate the gun is “waking up”. Whenever the gun points in another direction, it is basically moving it’s head. This could be used in a humorous way.

These are all little ideas, but they can add up and make a sterile and cold video more engaging and fun to watch. Do you have any ideas you’ve thought up from seeing this video test?

Atom - Thanks! I appreciate the encouragement! I’m hoping adding a “look development” phase to my workflow will produce better end results.

Frobenius - Thank you! One of the benefits to making these “all CG” videos is that you have no limitations with the camera or lighting or materials; you have all the control in the world, and money isn’t so much of an issue. Of course, time is an issue, and you have to keep things in perspective and know when you’re spending too much time on one piece of the puzzle. Your rocket engine tracking thread is coming along great, I’m enjoying watching it!

minoribus - I agree with you, I think this is better suited for the WIP section. I also decided to create a thread for it because I realized the sketchbooks section doesn’t get nearly as much traffic as this one does!

I was confused by that big, rectangular area on the side grip (or perhaps heatsink) part of the gun; there are cetainly limitations to the script, and the size and shape of different objects and areas of the mesh will influence the end result. There are a handful of settings that can be tweaked, which the creator of the script explains here, at this exact part of the video:

Are you can see, the script basically just uses Ambient Occlusion and inverted AO to define edges and cavities, and “neutral” areas, so that is why it is a good approximation, but not always perfect. In fact, I’m not even using the generated dirtmaps to their full potential - I’m only using them as factor inputs. I’ve not been able to find a .blend file which uses Nicolò’s script with proper material node setups, so I don’t understand how he is able to use the middle gray values in his dirtmaps to correctly define the base material areas. I think for this project, the gun looks “good enough”, so I will focus on other things, but in the future I’d like to learn how to use the dirtmaps to their full potential!

Nickersf - I’m glad the links may help you! I’m looking forward to seeing your sketchbook!

I’ve seen some works created with ddo2, and on multiple occasions have looked it up, but the official tutorials have not been easy for me to understand after the first 5 minutes (maybe because I have no experience with the interface). Would you recommend it? Also, do you have to supply all the textures it uses, or does it come with some? In one video, I saw that it created areas of chipping, water lines, and other interesting looking “grungy” sorts of texture. Is that an internal parameter that can be tweaked?

As I mentioned above, the dirtmap script works essentially like AO does, so your results may be a bit smooth. One trick I found is to take the dirtmap you’re using for a specific input (say, your edges material) and multiply it with another image texture of grime or scratches, whatever works. In my case, I’m using the white areas of each dirtmap as a factor for a mix shader. So this multiplied result of two images will be more random and less smooth than simply using the dirtmap by itself. I can upload an example of my current node setup if my description is too confusing! I’m using an older version of Blender (2.66 I believe); I found when I tried rendering the model in 2.69, by bump maps looked completely different (pretty much gone), so I’ll be sticking with my older version for this project.

Thanks for the long explanation, James, very appreciated. I understand that it is important to reduce the developing of models and textures to a reasonable amount in order to produce something in time. And of course we in the BA forums do look with a special eye on such things. Besides that the gun looks great and as I already said the dirtmap script seems to be very promising and you’ve already produced a good result with it.

In your look test I see some problems with the glossy material of the gun. There is some flickering in all the areas were it has the pure glossy material. That’s always a problem for me also, when I render animations with cycles. The glossy areas need extremely much samples to clear. At some point I normally say, that I have to live with it, but you are doing closeups. Have you tried filter glossy? And there is a new option for the clamping of indirect light in 2.7, that might help. But I know that you want to stick with 2.66 because of the bump map. Perhaps it could also help to use a different seed per frame. To do this you can write a “#” into the seed setting at the render panel. This then uses the frame number as a seed, thus giving more entropy to the distribution of the noise.

I had not heard of this clamping option in Blender 2.7, thank you for mentioning it! I may have to try some sort of filtering to reduce fireflies, but I’m not sure yet if it will be necessary. My look test was rendered with 64 samples per pixel, so when I increase it to a higher quality, the flickering may be reduced. However, I am using a small sun light in the scene, which doesn’t help fireflies. I may try increasing the size of the sun lamp, even though I will lose the hard shadows.

And thank you for the #frame tip; I’ve been using it for a while now, it really does help make the noise look more organic and less conspicuous. I almost wish it was set that way by default, rather than having to remember to set it up every time.

I’m working on a hatch door today, I’ll have some images of it to show soon. I may also create a storyboard for this project, to aid in telling the “story” more effectively.

In an unrelated note, someone I know posted images of a very cool looking physical model from the Titanfall game. I’m hoping they will let me come visit them so that I may take images of it to create a 3D mesh for use in a future video.

Digging that grunge so much! Looks really realistic.

Looking at the video in post 8, I can see everything is going right with this project. Man, that is rendered beautifully. The process you’re using to make this is fascinating and I look forward to learning more.

Very very nice James. And thanks too for explanation, very appreciated. Nothing to say :wink:

Goldenfrog - Thanks! I’m hoping the final product will look very realistic, I think it depends mostly on my materials and lighting. Still working on it.

Frobenius - Thank you, I do feel that incorporating “Look Dev” tests into the early stages has helped me a lot. I’ll try doing this more often.

ookka - It’s good to see you stopping by! I admire you attention to detail, and so appreciate your words very much. I’ve been studying more tutorials, and have learned how to better use the dirtmaps for better effect. I’ll be sure to share my testing soon.

Today I’m beginning the process of PhotoScan mesh generation. With nearly 400 images, it will take some time to complete, but I’m confident the results will be worth the wait. I’ll share test renders of the environment later today as they become available!

I’ve also made a change to the type of targets the gun will be shooting. Originally, I wanted to have typical wooden targets, to really show the gun hammering away and shredding them to bits. However, this would require a lot of additional work to get a good looking effect, and with my time constraints and the amount of work left in other parts of production (not to mention rendering it all out), I’ve decided to make the targets holograms/holographs instead. It may not be as satisfying as seeing the gun tear through a wooden target, but it’s a heck of a lot easier to achieve. Plus, the focus of the video is going to be the gun anyway, so I think it’s an acceptable substitute.

I just wanted to share that decision with everyone, because it’s an example of some of the types of choices you have to (reluctantly sometimes) make when working on something. It took me a long time and lots of late projects to realize it’s important to make those decisions as soon as you realize an issue exists.

Here’s today’s update: It’s 4:30 in the morning, and I’ve been able to work on the assets throughout the day. Working on the shader for the bullet shells, I can’t decide if they should be shiny or dull. What do you think?


or Dull:

The shells all have the same texture, blended box mapped onto the model, so you can see the texture mirroring up close. I’ll make sure to give them better texturing for the up-close shots, but I doubt anyone would be able to tell in the majority of the shots.

I’ve also been working on the hatch that the gun rises up out of. This may look pretty basic, but I’m pretty abysmal at modeling, so this is world class work as far as I’m concerned:

There’s still a bit of work to do on the hatch and hologram mechanisms, but that’s for tomorrow. A lot of the believability is reliant on the texturing and shading, so the modeling is only half the work. That said, I’m proud of the hatch so far, and will be uploading it to Blend Swap once the film is completed.

The gun throwing the shells is shot from very interesting perspectives. I like especially the last one, where the shells are flying towards the viewer. The hatch is well modelled. The backdrop in the second render is a placeholder, right? These building are too much of a contrast to the scifi weapon :smiley:

With motion blur and with the speed of the shells, the box mapped texture shouldn’t be a problem at all.

It’s 4:30 in the morning

This project has caught you, I believe :slight_smile:

Good ol Hammerburst. So is this gonna be a bunch of different iconic guns from various movies and games or is the Hammerburst just a placeholder?

minoribus - Yes, I try to find interesting camera angles to make the film more interesting. Because the gun is serving as the only “character” in the video, I want it to feel larger than life. So a lot of times I’m having the camera placed low, looking up at the gun, and trying to fill the frame with it. And you’re right, the buildings in the background won’t be visible in the final work; they’re just part of the HDR map I’m using at the moment. I’m not settled on the final lighting setup yet, but I got this HDR background (and a couple others) for free from Ten24 about two months ago. They’re pretty good, and might still be available for download.

xalener - The Hammerburst will be the only gun we see in the film, but it’s funny you mention that; my friend John (from the GTA and other videos on our channel) said we should make a video showing a group of people test firing a bunch of iconic guns from different videos games. We made a list of about 26 different games that had great, memorable guns in them, and will be trying to cram that video into our lineup before the end of this year. We may change the format and make it like a deathmatch or something else, instead of a firing range, but I think it could be a really great idea. We just have to choose which guns will make the cut, then build them (in real life) before we can shoot the movie. A lot of work to be sure, but I’m excited about the prospect.

Today I’ve finished just about all the modeling of the hatch (whoo-hoo!), so tomorrow I’ll begin arranging all the pieces that just need to be duplicated and positioned, and setting up the constraints to control the opening of it. Then once I know it will all work properly together, the texture creation process can begin! I think the texturing should only take a couple days.

It’s about 4:30 in the morning again. Update pictures coming soon!

Hey James…it’s looking great and personally I prefer the duller shells…but why do they have a hole in the back? Looking forward seeing the new photos can data! This is going to be a good one!

Thanks, Shakedown! I prefer the duller bullets as well, I feel they match the look of the gun better. The holes in the back of the shells are where the primers were. I removed them for purposes of testing rigid body dynamics (just to have fewer polygons taking up memory with thousands of shells), but I’ll be sure to pop them back into place! I’m using shells from a collection of bullets JackBryanReynard uploaded to Blend Swap: here.