Blood, Sweat and Code
by Netwyrm · 11/09/2015 (5:13 pm) · 4 comments
Made it through the first phase of beta, from 10/16 to 10/31. Phase 1 consisted of a pumpkin hunt where collected pumpkins were redeemable for costume part masks, and scarecrow demon spawns were placed at chokepoints on all the maps. The scarecrows had their own set of masks which awarded randomly. And now working on phase 2.
One of the Halloween masks players earned through collecting pumpkins
Not much uptake, actually, but all the systems written to support everything outside of the game worked unassisted, and that was the real point of the first phase of the test. I don't think a game really exists outside the infrastructure which allows for signups, automated email confirmations, players changing their addresses and so on--and especially not without key management and key redemption in place. Not the most exciting part of the project, but it's really, really important, especially when you take the one-to-many nature of indie development into account... in order to scale at all a large part of the day-to-day customer handling has to be automated.
Had an introductory manual printed up, made some fly-in cards for fun (and to have something to stick the beta key to!) Made up a template so mailing labels could be printed directly off the web, and everyone who signed up for beta got sent a beta pack if they were in USPS territory, and a PDF of the manual with a serial key via email if not.
Overall, I'm pretty happy with the way it ran... would have liked more beta players, but am only too aware self-promotion isn't my main strength. I'm especially happy that the game's infrastructure is complete enough that I was able to set up the Halloween event in less than a week using the backend tools I'd written, including setting up 70 spawn points for the scarecrows across 11 maps.
Phase 2 will run from 11/16 to 11/30, with signup at www.worldofantra.com, but I am running a bit slow. Maybe it's the time change, or maybe charging back up from what it took to get phase 1 out the door. Trying to get at least one of the three tiers of housing working in minimal fashion, but that's tied into the crafting system so intimately that both pieces have to be done back to back in mirrored fashion to pull it off.
Other issues remain--especially optimization. I have the weirdest issues with frame rate... there's a spot in one zone where everybody reports a drop to 4 fps, then there are spots which affect some players and not others. But then there are zones with the same models and materials and much more besides, and they work more smoothly. Head scratching ensues!
Even for me, that's a record oO
Above is a screen capture showing a truly ridiculous number of draw calls--which I'm pleased to report after this weekend's massive crunch on model work are down to 3,000 in the same spot--still not great, but a good deal closer to reasonable.
And, just for fun, here's one from this morning before I got into scripting, at a glassblower's stall in the marketplace.
Can't collect water from magic pools without something to carry it in
And I wanted to say thanks, Sorin, Steve, Sean and Dwarf King, for posting comments to my last thread. It really helps to log in and see those words. Dwarf, I did pull my old dumbbells out and have been tossing them over my head every day. Got pretty tired out there, the week after making the beta date is a complete fog, so yes getting some exercise is not a bad suggestion and I thank you for making it.
One of the Halloween masks players earned through collecting pumpkinsNot much uptake, actually, but all the systems written to support everything outside of the game worked unassisted, and that was the real point of the first phase of the test. I don't think a game really exists outside the infrastructure which allows for signups, automated email confirmations, players changing their addresses and so on--and especially not without key management and key redemption in place. Not the most exciting part of the project, but it's really, really important, especially when you take the one-to-many nature of indie development into account... in order to scale at all a large part of the day-to-day customer handling has to be automated.
Had an introductory manual printed up, made some fly-in cards for fun (and to have something to stick the beta key to!) Made up a template so mailing labels could be printed directly off the web, and everyone who signed up for beta got sent a beta pack if they were in USPS territory, and a PDF of the manual with a serial key via email if not.
Overall, I'm pretty happy with the way it ran... would have liked more beta players, but am only too aware self-promotion isn't my main strength. I'm especially happy that the game's infrastructure is complete enough that I was able to set up the Halloween event in less than a week using the backend tools I'd written, including setting up 70 spawn points for the scarecrows across 11 maps.
Phase 2 will run from 11/16 to 11/30, with signup at www.worldofantra.com, but I am running a bit slow. Maybe it's the time change, or maybe charging back up from what it took to get phase 1 out the door. Trying to get at least one of the three tiers of housing working in minimal fashion, but that's tied into the crafting system so intimately that both pieces have to be done back to back in mirrored fashion to pull it off.
Other issues remain--especially optimization. I have the weirdest issues with frame rate... there's a spot in one zone where everybody reports a drop to 4 fps, then there are spots which affect some players and not others. But then there are zones with the same models and materials and much more besides, and they work more smoothly. Head scratching ensues!
Even for me, that's a record oOAbove is a screen capture showing a truly ridiculous number of draw calls--which I'm pleased to report after this weekend's massive crunch on model work are down to 3,000 in the same spot--still not great, but a good deal closer to reasonable.
And, just for fun, here's one from this morning before I got into scripting, at a glassblower's stall in the marketplace.
Can't collect water from magic pools without something to carry it inAnd I wanted to say thanks, Sorin, Steve, Sean and Dwarf King, for posting comments to my last thread. It really helps to log in and see those words. Dwarf, I did pull my old dumbbells out and have been tossing them over my head every day. Got pretty tired out there, the week after making the beta date is a complete fog, so yes getting some exercise is not a bad suggestion and I thank you for making it.
About the author
My adventures in T3D are chronicled at http://www.worldofantra.com. Please be aware the subject is sword-and-sorcery, and the occasional bloody or bare body part may be in scope.
#2
I am using zones and portals for the interiors. I pick up three or four fps when inside my latest interiors by doing so.
I rebuild models, every time I learn something new or figure something out. They are pretty badly structured, with many too many polygons. I was laser-focused on draw calls, because that's what I've seen talked about here on the forums, but after reducing and reducing materials, I realized it was only part of the equation and am now going after polys and primitives.
A mistake I made was to use a building tool which made it easy to construct things but created enormous numbers of split faces in doing so, and then proceeding to build insane things with it.
In the latest go-round I've made sure to build the model in parts, so each piece can be cleanly separated and tested independently, then reassembled in the engine. My latest models are much better technically, using two shells--1473 faces total as opposed to 16332 faces in the models they are replacing.
I'm also using extremely aggressive LOD with the newest buildings, so the interior portion does not render unless you are right up upon or inside it, only the outer shell. This also lets me style the outer shells to provide some variety, without having to touch the interior geometry once it is all solid.
I chalk it up to experience, or rather the lack thereof.
11/10/2015 (8:54 am)
Hi Dwarf King-I am using zones and portals for the interiors. I pick up three or four fps when inside my latest interiors by doing so.
I rebuild models, every time I learn something new or figure something out. They are pretty badly structured, with many too many polygons. I was laser-focused on draw calls, because that's what I've seen talked about here on the forums, but after reducing and reducing materials, I realized it was only part of the equation and am now going after polys and primitives.
A mistake I made was to use a building tool which made it easy to construct things but created enormous numbers of split faces in doing so, and then proceeding to build insane things with it.
In the latest go-round I've made sure to build the model in parts, so each piece can be cleanly separated and tested independently, then reassembled in the engine. My latest models are much better technically, using two shells--1473 faces total as opposed to 16332 faces in the models they are replacing.
I'm also using extremely aggressive LOD with the newest buildings, so the interior portion does not render unless you are right up upon or inside it, only the outer shell. This also lets me style the outer shells to provide some variety, without having to touch the interior geometry once it is all solid.
I chalk it up to experience, or rather the lack thereof.
#3
11/10/2015 (7:35 pm)
Nothing informative to add, but great work! 
#4
About performance:
1.use zones and portals if ever possible.
1.a. also culling volumes are good to hide a lot stuff to get rendered.
2. use up to 5 LOD steps
3. Most important tip:
Use when ever possible only large atlas textures for a whole asset.
Example:
make one large atlas texture for all buildings, one for all foliages, one for all trees, one for all characters, one for all animals.
That way you have low draw calls!!!
Draw calls is the most performance killer in t3d.
I made a 15 x 15 km world with this atlas texturing system, and i get average 30-40 fps. Also when a lot creatures are around, or a lot trees, fps is nice.
Alot polygons above 3 million in the scene, also cut the fps down.
So really strict using only atlas texturing, is the way to go, in t3d, for getting nice performance in larger scenes/worlds.
11/12/2015 (3:35 pm)
This all remember at a game i play! Nice work!About performance:
1.use zones and portals if ever possible.
1.a. also culling volumes are good to hide a lot stuff to get rendered.
2. use up to 5 LOD steps
3. Most important tip:
Use when ever possible only large atlas textures for a whole asset.
Example:
make one large atlas texture for all buildings, one for all foliages, one for all trees, one for all characters, one for all animals.
That way you have low draw calls!!!
Draw calls is the most performance killer in t3d.
I made a 15 x 15 km world with this atlas texturing system, and i get average 30-40 fps. Also when a lot creatures are around, or a lot trees, fps is nice.
Alot polygons above 3 million in the scene, also cut the fps down.
So really strict using only atlas texturing, is the way to go, in t3d, for getting nice performance in larger scenes/worlds.
Torque Owner Dwarf King
Noble Games Production
Do you use zones and portals?
You can use them inside buildings so when the players are not inside them the engine will not render them. If you have many rooms and small shapes in the area you could squeeze better performance out with that technique.
The portals are good for the doors so players can look inside the buildings when in the right angle.