[Nils]

A Christian 3d adventure in development

The plan was to start in March with the further development of this project after a long pause. When finally all was ready to go, the Covid-thing started here in China (February) While taking care of the family and chronic illness was there barely any time to do work, sometimes here and there a little. Now almost everything is getting back to normal (a new normal that is) I'm looking forward to spend some full-time months on this. After a year it's still in the starting phase, but not to worry; there are no strict deadlines... The engine is already much outdated, a few more months really doesn't matter.

The aim is to target old hardware; it's being developed on a 6 year old low-end game PC and will be satisfied if it keeps running with a 30+ fps at 1080p. Realism is not a goal, but just looking decent enough is something try to spend some time with.

Hopefully, when I'm managing to get some help on the coding side, some upgrades can be made in the near future. The current engine branched of from 3.6 and many changes have been made. Porting to 3.10 or 4.0 is a big step, but will see how it goes; in the meantime you can enjoy the D3D9 renders below

Please remember that everything is in a early stage of development, and all may be subject to changes; nothing is fixed yet !

























Previous blogs:
June 2019 viewtopic.php?f=8&t=1654
November 2019 viewtopic.php?f=8&t=1678
and viewtopic.php?f=8&t=1704

Work in progress, unfinished work, not final yet, will change in the near future, not everything is done yet !

[Nils]

In this video you see the quality settings function being implemented, though the actual settings are far from done.

Watch it on YouTube (FullHD1080p): https://youtu.be/f2f2upfmvHg

[Nils]

Sounds are also a work in progress

Watch it on YouTube (FullHD1080p): https://youtu.be/Yl1kqdjByv0

[Duion]

Looks very realistic and finished to me, hard to tell if there is something you still can improve, definitely one of the best graphics ever made with Torque.

The only thing I wonder is, if you made all the models yourself, since there are quite a lot of different scene objects, even buildings with interiors.

[Nils]

Thanks @ Duion. It's nothing comparable with the current standards though. But I'm really not aiming towards that. I'm happy with balanced scenes where it all matches together.

I hardly buy assets. There are some I bought a while back like the vehicle (Lada) and the rowing boat for example; but the conversion to a decent working and T3D asset (texturing, LOD's etc.) takes so much time that the consideration to buy something often leads to not doing it or not using it. The latest 3d model I bought is a blue whale on TurboSquid. To model something like that animated and all would take me ages. All buildings and structures are done in-house, as about 90% of the props.

This is the latest building I'm working on, need to finish the tower and texturing.



The switch from Max to Blender takes so much time (sigh)

[Duion]

In art there is no such thing as "current" standards, either something looks aesthetic or not.

Sure today things have more polygons, higher resolution textures etc, but this all does not help you much, when you don't have the skills to assemble them in a realistic way.

I see it in a lot of "popular" games where the "artists" basically do almost everything wrong you can do wrong in their level design. Sure they have a lot of high quality standard assets, since they buy them, but then they just take them and randomly place them on the terrain, making no distinction between biomes or context. They have a meadow, then place some trees on it, some houses, some rocks, some roads and call it done, never considering that trees grow in forests and houses are in the city and rocks are in the mountains.

Regarding assets, I also found that I had to make almost everything myself, since the mere process of converting and fixing them for my purpose is almost as much work as doing them myself from the beginning. But some assets like the car for example I would never bother to make just to have one of it as a scene prop.

[Nils]

The protagonist (Johnas) has a new body & face (including all new animations). Still working on the clothing though. The choice to stick with a bit alpha for the hair & facial hair has to do with performance. Eventually you'd like to pack as much as possible to reduce the draw calls and unused spaces in the atlas maps; Combining lerAlpha with other materials is not working properly with this 3.5 fork of T3D so even does not leave much choice.



Porting this custom clone to a recent branch may be not feasible; The linear lighting of the terrain materials in the DX11 versions will break every bit of diffuse map I created for the terrain, as well the tuning of the colors of vegetation & rocks to the terrain will be all in vain. Unless a solution is presented may I be continuing with D3D9. A custom shader perhaps? The point is that there are currently no/not much developers with the same issues I believe.



At this point is the game much CPU bound after a week of improving performance on the GPU side (reducing draw calls, higher/more compression etc.) On average TMF is running with 1920 * 1080px between 30fps and 40fps with occasionally a drop to to 25fps with dense scenes that still need to be optimized on a 5 year old rig (i7-4770 3.50GHz CPU / 16GB RAM / 970 GTX - 4GB GPU / 1080p - 60Hz monitor). That's with AA, VSync and Anis filtering on.

[Duion]

You may be able to upgrade to 3.10 or whatever the latest full release was, yes there will be some lighting settings you may have to adjust, but it is doable, the terrain texture system was changed at one point, but when I pointed out how bad it was, it was switched back to the old blending algorithm.
Version 4 which is not released yet will break everything again, so it may be wise to not update to it, not even when it will be released in a final version, since in my experience even final releases are not production ready, because hardly anyone is testing the versions in an in production game like you have. Which means you should upgrade and test it, then it will help the development of Torque, but otherwise it will be much more work for you.

Regarding performance in my game I reduced the terrain LODing by a lot, even less then the default minimum setting, it is hardly visible and saves you a lot of polygons.

Regarding performance however the biggest offenders are lighting and postFX, the other settings hardly even matter. In my big outdoor level I had to disable SSAO, because with lots of vegetation SSAO really kills performance. Parallax on terrain I disabled alltogether, because it is really bad for performance and does not give you much.

Your shadows are very sharp, so you may be able to reduce the shadow resolution as well as the shadow update rate, but i think the shadow caching feature was implemented later, maybe in 3.9 or 3.10, but according to my tests it does not give much performance either.

I have a similar rig but with a geforce 1050 card, the cheapest of the new generation and my game runs on 60 Frames on max settings, sometimes 50., that is what I'm aiming for. Of course if you have 50 bots or players running it will reduce to critical 30 FPS. But 30 FPS for a non fast paced multiplayer game is okay on a low end gaming rig.

[Nils]

Thank you for your response Duion,

Graphics settings in this game are fully functional. Things like Terrain LOD, shadow detail etc are all accessible for the player to tweak to their system performance/preference. The medium settings are targeted for my 5 year old rig tor run with 30fps on 1080p: which is good enough for a free adventure game. The screenshots you see are mostly made with high settings (the last 2 are even 4k)



Stuff like AA and SSAO and PostFx do not have a significant/noticeable hit on performance in TMF. Things like parallax aren't even used at all; the first choice for texture format is DXT1/BC3 and may choose "better" compression when it's simply not acceptable to look at

[Duion]

While you as a developer think your option menu is great, because it allows users to fine tune every detail, it may actually be bad, because you should not give the user too much freedom of choice, because they will get confused and screw it up. From my experience many users do not even bother with the options at all, so your best choice is having good default settings and then the options should be simplified, like low, medium, high, ultra or so. My initial plan as also to expose every variable to the end user, but I ended up with a low, medium, high system except for postFX and decal lifetime.

Just having a bar with no numbers is a hard indicator, most people will just slam it to the right and then complain because the performance is bad. Trust me I speak from experience. But I see you have a basic and advanced tab anyway. You may even can get away with hardly any options at all, since that is the trend in video games now, as little as possible options. I already played games that had no graphic options at all.

So you will make the game free? I would suggest making it small priced, since free games can be good, but also can be bad, it is a hard decission. I experienced it myself, as a free game, I was cut off from some methods of getting visibility, like having reduced price or promotion from free key giveaways. You have to carefully plan your launch, that is more imporant than the quality or performance of your game.