June 17,2014

OK was not realllly ready to do an update just yet but I just got my new GeForce 780ti up and running the way I want it and and finally got around to getting ShadowPlay working.



To let the cat out of the bag, as hinted already, Nevada is being rebuilt for UE4 which is a redo of Turnpike. Nevada is a map redo by Skaz originally intended for UrT/fstech1 which sad to say is a dead end project but I still have all of the major mesh components already done during the brush>mesh conversion intended for the fstech1 design.

Even with physics based materials it's a fast map to play but at the moment it is the shortage of ready made materials that is slowing things down as far as a quick conversion goes. We do expect this to be resolved once Epic has the Marketplace set up and over time collect as many different, fair use, materials we can get our hands on or make ourselves.

Back to the blog in hand.

The good news is I/we can finally do real in game captures of work in progress to give a much better scene as to use of space as well as speed that just can not be done with out hardware capturing. Up to now I've been using software capture that although works it's not going to really be all that impressive as to what we are seeing on our end.

Sooo for now on full resolution, 60 fps game play, with sound.

The bad news?

This stuff is to cool for school so expect yet more work in progress not fit to play. ;)

As an added bonus I also added a run around, maybe to fast?, and yet more work being done on the new sentry gun.

One is the model design to be used, yet to be fully textured, and the other is a mockup of the AI done in Blueprint.

Here is the second part, with adrenaline shots, akimbo weapon mode, knife & handgun combinaison, and a new way to curb stomp. These are suggestions and i think finally of continuing these boards to add Silverfox suggestion.

Needing a break from working on JUST Casa I've begun work on the Mr. Sentry replacement and I've completed the bulk of the modeling work and the AI base and now in the process of doing the layered UV mapping and texture work.



If you don't know who or what is a Mr. Sentry it's as you can see your basic mini-gun design but is a fundamental map object as part of the CTF game type. It's usually placed inside the team base to protect the area from the the opposition from gaining a foot hold of the spawn area and more or less run the capture count up unfairly.

Need more info feel free to post questions in the Urban Terror HD forums.

In animation you have the called key poses that are the most important poses of a movement. They are here to express the movements. Then you or the assistant usually draw all the poses in between. Even if you can animate in straight, meaning you draw one pose after another until the end, this Method is usually the mostly used. With this idea in mind, i've tried to apply a bit of this method for Urban Terror 's moves when FrankieV started the animations. I've hesitated sine this is from last summer, but here are some of the ideas which are not that different from what we know.

I think it was important to resume the main moves of urt with pictures.

So making dynamic pose in Urt is more restricted than in animation since there's a gameplay to respect. It's not about drawing an in-your-face pose but more something close to 4.x with a character not throwing his arms everywhere during a climbing. Of course regarding the movement, the pose can be more or less "complicated" in my opinion. Climbing poses for me can be a bit more in-your -face since it happens while the player is almost static, while a wall-jumping animation (althought a bit exagerated here) has to be simple & dynamic, with the shortest transitions.

It can be messy as sometimes you got a 1 > 2 > 4 > 3. You can have two options for a movement. (follow the arrows).

So here we go, yet another blog. To be honnest I'm not drawing that much for Urban Terror anymore. I've started to model props for our maps instead. So it can be used by any mappers. I'm trying to contribute to the props bank so we'll be more in a "what can i take from here", rather than a "could you model me one thing". Since a simple props can inspire a bit of a part of a map. For now, it's just about having the most common stuff you'll find in any streets basically. I'm pretty sure that in the environments we want to build, most of these props will find a place. The way it's supposed to be used, as trashes, or gameplay elements.

This new hobby plus Frankie waiting for me to modify 2-3 things on Ghosttown to convert it so i can have fun with UDK as well, I don't think it'll be usefull to draw for URT at all today. I'm not at the texturing part yet. I'm just trying to chain props modeling as much as possible for now. So nothing fancy at all to be honnest, but our environment need all that kind of small stuff to look good. Ghosttown converted soon will be a good place to test these models. Also, I plan to work with Blade on modeling different parts of a characters, simple things like pouches, holsters, glasses, berets, gloves, anything characters related.

So here is what will be my 2014's route.

I'm still wondering what can I show here while staying coherent so for now, I got some old drawings from last summer that are still usefull, despite all the recent changes. It concerns characters animations, revised & refreshed a bit plus new ideas which are not a priority for now but extra propositions. I'll post a first part of these boards later today.

June 8,2014

Yeah a bit soon for an update and as far as what has been done up to this point is I worked a bit more on the landscaping to get a bit more texture in to it as well a bit more interesting shape than just a flat surface, mostly, as well add a few more map objects and buildings to help out with the parallax.



The big change though is not what was added to to the map but what Epic added to the release of UE 4.2 , that makes the effort to do yet again another post so soon, and a few key additions to their up data is to have meaningful run time stats that includes an on screen FPS stat so moving forward with my walk about's I'll be including that bit of info.

An interesting bit of info that I ran across, and still need to look into, is apparently EU4 can be performance crippled so testing can be done as a simulation of lower end specs so if that turns out to be the case we can then determine what the minimum specifications will be for the HD release.

What is of interest though is the Mac is finally uping it's game when it comes to gaming with the introduction at WWDC of the Metal API for IOS 8 with rumors that it will be included in the future in UE4.

https://forums.unrealengine.com/showthread.php?7946-UE4-%E2%80%9CZen-Garden%E2%80%9D-using-Metal-API-for-iOS-8-revealed-at-WWDC

Kind of makes you think will the Mac become the preferred gaming rig choice of the future?

Also for thous asking about the player animations, as I know that’s somehow become a sticking point in UrT 4.2, I plan on doing a blog as to how things will be set up using visual aids.

June 4,2014

For this update I've add a bit more detail into the geometry, added trimmings and gutters, as well as further iron works and some foliage to get a better sense as to how well the use of space is building up towards the final result. The final result of course will be totally different than what’s being represented as part of the concept stage as the act of changing one thing usually means having to change another as to fit to finish.

Also to move things along a bit further as to lighting and shadows I made some adjustments, added a few entities, towards creating a balanced and linear lighting solution that also requires balancing out materials along the way.



Normally this is a pass best left for last as you can find yourself putting in to much time and effort doing the same requirements over and over again but out of the box UE4 is not really set up to give a quality render even as a working solution.

There are are a lot of features in Lightmass that just does not work with out some entity additions that needs to be added or adjusted as secondary requirements to added light elements. Perhaps the single most important parameter as part of the Lightmass World Setting is Diffused Boost that if left at it's stock setting of 1 none of the advanced setting will be effective if they work to expectations at all.

The nature of natural lights is they just don't hit a surface and go splat but rather bounce off the surface based on it's reflective index and when it does so carries along some of the color information along with it and transfers that color to surrounding surfaces and in the process picks up further color information in the next bounce. What you land up with is a hue of different light pooling that if balanced just right looks at the very least nature even if not realistic.

More or less this is what Diffuse Boost does that allows every thing else to work as expected like Ambient Occlusion that by it's self will look like a painted edge with no form of connectivity between other surfaces with in the same area.

The next secondary need is for some form of reflection capture as all materials should have some level of reflectivity index as lightmass bounces light into a given area. To accomplish this requirement a reflection capture needs to take place with in proximity of connecting surfaces and in some cases overlap as to the desired effect. In other words if it looks good then it's good.

It's interesting how reflection capture is preformed to it's real wold counter part by the use of a reflective object, by sphere or box, that suggests that an accurate lighting solution can not be achieved by the use of a single light entity but by the addition of secondary elements that supports the desired result based on observation.

It should also be noted that surfaces that are highly reflective, like metal for example, will not work at all with out a reflection capture and the result will be a black surface render of a surface that is not with in direct line of the light source.

The last thing of note that goes hand in hand with lighting is shadows and although there as somethings that directly or indirectly effects the overall quality the ideals of density does not apply but is controlled by the overall Environment Color used to balance the fusing of the light solution based on other parameter changes like Diffused Boost.

With out a doubt a lot more complex than one is use to as to map lighting for idtech3 but as part of a process it's best practice to focus first on use of space best suited to the ideals of someone moving through an area as well as to what areas a player can or can not be allowed to enter and leave the eye candy for the polish phase.

May 26,2014

OK in this update of Making Casa I'm going to cover a few questions popping up.

Looks to blocky.

Yes it does as 99% of the buildings are directly converted from brushes and more or less they are just box meshes that do not take well to advance lighting and rendereing system, like the one in UE4. Most problematic is smoothing groups, light, and shadows as they do not like right angles greater than 45 degrees and the result tends to be a hard knife edge corner.

There is however a simple solution to the problem that by chamfering a right angle edge, even by a small amount, the corner now becomes 45+45 and now Sg's, lights, and shadows will create a much softer transition.



Need tessellation.

The ideal for Casa, as a map that will come with the game, is to keep things as light as possible to maintain a high performance envelope to keep the competitive community happy.

Granted it's all going to look a bit dated but I'm sure third party will come up with the eye candy.

Lighting needs improving.

Good lighting is not a factor of lighting alone but an interaction between, lights, shadows, as well as texture levels that are properly balanced. The best practice is to stick with a working level of lights and as part of the polishing pass balance texture levels to match the environment.

Note: This is a series of videos of Casa a, well known map for the game Urban Terror, showing work in progress as it's converted from idtech3 to Unreal Engine 4.



In this milestone video all of the cheats of photo sourced windows and doors have bee replaced with geometry as well as balconies added as map objects. A proper sky dome was also added to help improve over all lighting and shadows.

Some things of interest the facades of the buildings themselves were converted directly from brush to polygon by exporting them from the original .map file and the total face count for all of the buildings is less than 10k. By comparison HD weapons will be between 7-10k with player models between 15-20k, give or take based on the design, so the actual filling of a volume of space is highly efficient as compared to doing the same environment over again using brushes.

Also of interest is how much more efficient UE4 is when it comes to map objects in general and although there is a lot of windows and door casings they are all modified versions of a single window object that has been instanced copied so it's memory foot print is equal to the single object.

In all though if not for the less than ideal resolution of the texture images themselves a 1-1 conversion of a brush based .map map could easily be completed in a weekend and be considered done however the hard part creating a mesh based map is the need to have in place a proper source control as the map grows using progressive design ideals.

Source control for Casa was a “MUST” as what was extracted from the .map file was a massive amount of mesh and texture data that in it's raw state was total unmanageable lacking any kind of logical separation between supporting sub-components. A way to think of it is being lost in the forest when your job is to cut down the tree. What tree?

The first iteration of attempt to create a usable source control chain started during testing in UDK as to design pathways that best fits our needs as well as to have in hand a visual representation as to performance differences switching from idtech3 to UE4.

At the time we switched from UDK to UE4 it could be said that we had the working structure for HD done, with the exception of movement mechanics as we were able to run around in game using Casa with our weapons as well as our player models and it was clear as a bell the a game does not inherent engine characteristics but are added by design choices.

Back to the subject in hand in UDK the base work for source control was achieved to some degree by breaking things down into manageable blocks and each block given an identifier and exported relative to world origin 0 0 0. Doing so is not really ideal as to map objects in general but the result is each block would snap into place like a perfectly fitting Lego block.

Anyways to keep things short this is enough for now, with more to come, with a lot more divide and conquer in the mix as we have yet to disuse the pain of redoing the textures.

So it's all very well and good this content delivery network stuff. However it's super premium service which costs a bomb to deploy globally. When you run all day every day at near 400 MBit/s the Bandwidth used just racks its self up past the scope of you budget.



The way going forward (i will be replacing our current download providing) is to have enough physical capasity to offer the day to day stuff with CDN ready to go for big deployments and to soak up that "Extra Load". I'm sure I will blog about that soon... for now... back to it!