[QUOTE=JWki;51992258]Was that Sean Parent's talk at GoingNative '13?[/QUOTE]
Yeah, and I have never made it through the whole thing. Didn't manage to do it today, either.
I make it through a chunk, learn something new, then run off to go apply that thing and forget to come back :v
[editline]edit[/editline]
the tldr of it seems to be that the algorithms library has everything one could ever need, and that learning to recognize common-use algorithms in your code can help simplify your code (and improve it in many cases, since these algorithms are probably more stable than your code)
*ruff* What's that boy? *ruff ruff* I've done more hand animation work? *ruff*
Problem: Everything on a character model is the same colour. Reddy red red. Its very difficult to see whats going on with the hands because they're just red like everything else
Potential solutions:
1. Use a real bone rigged hand model. Pfft. Although really this is a no because I don't think it'd fit, I'm unwilling to completely change the art style (because then I'd have to use proper models for everything), and also because moving form more procedural -> proper modelling means I'd have to get familiar with blender and probably more traditional animation techniques when I update the main character, which is going to be rather a lot of work. I would like to finish making this game eventually
2. Nick part of a hand model and fit it into the art style. I had a go at this one
Deconstructed and chopped down hand model from a knight I bought
[IMG]https://i.imgur.com/veaYlOQ.png[/IMG]
Suffers from the problem of "this fixes nothing". You can kind of see the *hand* better than before
[IMG]https://i.imgur.com/B2rOxM0.png[/IMG]
But really it still suffers from red on red friendly fire. IE it does nothing for the actual fingers, its still difficult to see what the hands are doing, or where they are
3. Remembered valve's talk about how you should bevel the fuck out of everything even just a little. While updating my finger cylinder model, I tightened the bevel significantly (and also chopped it down from 1500 tris to 350ish, wtf was wrong with me when I made it the first time??) which heightens the specular quite a bit on the corner, which makes it more visually distinct. This is difficult to see in a picture
[IMG]https://i.imgur.com/FBP3RjR.png[/IMG]
But it does make somewhat of a difference ingame. Its still not perfect though. I'll probably end up modelling some bits of metal to go over the fingers as well, and I may also apply a dirt texture to the hands (or sword) to clarify what's what
4. Actually modify the colours to not clash horribly. Still suffers from hand -> hand colour collision, but it'd be less bad. This does mean I'd have to come up with some sort of sensible colour scheme, and my artistry skills are absolute wank as anyone following my progress knows
5. Implement GI. Might seem odd but a lot of this would be fixed with proper occlusion and second light bounces which would give detail in lighting where there isn't detail in textures
[QUOTE=paindoc;51992678]Yeah, and I have never made it through the whole thing. Didn't manage to do it today, either.
I make it through a chunk, learn something new, then run off to go apply that thing and forget to come back :v
[editline]edit[/editline]
the tldr of it seems to be that the algorithms library has everything one could ever need, and that learning to recognize common-use algorithms in your code can help simplify your code (and improve it in many cases, since these algorithms are probably more stable than your code)[/QUOTE]
well fwiw the post inspired me to rewatch it today so someone did make it through.
yeah I don't quite agree with the no raw this no raw that on a global scope but the takeaway of knowing your algorithms and being able to select an existing algorithm for a fitting problem is pretty valuable.
[QUOTE=Icedshot;51992785]Hand stuffs[/QUOTE]
Try changing the hand colors a little bit, either darkening or lightening them up a bit. It should help with visibility. Doesn't have to be a lot, only just enough to make a difference.
I created my own implementation of Int32.Parse that averages ~700% better performance than the original method. It's missing a few sanity checks, which may slow it down slightly, but it's still quite a significant speedup. Smaller numbers can get as much as a ~2000% speedup.
[code]public static int Parse(string str)
{
unchecked
{
int n = 0;
int l = str.Length - 1;
for (int i = l, x = 1; i >= 0; i--, x *= 10)
{
switch (str[i])
{
case '-':
if (i == 0) return -n;
goto default;
case '0':
continue;
case '1':
n += x;
break;
case '2':
n += 2 * x;
break;
case '3':
n += 3 * x;
break;
case '4':
n += 4 * x;
break;
case '5':
n += 5 * x;
break;
case '6':
n += 6 * x;
break;
case '7':
n += 7 * x;
break;
case '8':
n += 8 * x;
break;
case '9':
n += 9 * x;
break;
default:
throw new ArgumentException($"The argument {nameof(str)} is not a number.");
}
}
return n;
}
}[/code]
[URL]https://gist.github.com/TheBerkin/decc733ac59fda3e344a09aa21d599f8[/URL]
A few things I've been working on lately.
Volumetric shader sandbox:
[t]http://i.imgur.com/GWlOxmY.png[/t]
[url]https://www.shadertoy.com/view/Ml3SD4[/url]
You can write your own volumetric shader in the "Buf A" tab.
Run-length encoded sprite / clown.glsl:
[t]http://i.imgur.com/rZjSs0S.png[/t]
[url]https://www.shadertoy.com/view/4tGGW1[/url]
Was messing around with sprite compression and decided to make something with it.
GPU Chip-8 emulator:
[t]http://i.imgur.com/t46Uc2X.png[/t]
[url]https://www.shadertoy.com/view/4ld3DM[/url]
You can change the rom in "Buf A" (recompile then reset) and you can get the view in the image by changing VIEW_MODE to DEBUG in the main shader.
If it runs really slow, try removing Buf A. It seems like some gpus can't handle the large arrays very well.
I think my game's combat system is finally ready!
You, the player, will be able to control up to 4 party members. Each party member has a key (default keys are H,J,K and L).
When your character is attacking, you gotta press their key whenever their weapon blinks in order to keep the combo going. You'll also have to pay attention to the arrows on top of your foe's head, because you have to press the movement keys accordingly. Failing to do so will result in weaker blows.
[video=youtube;NAt6Njiwx_I]https://www.youtube.com/watch?v=NAt6Njiwx_I[/video]
Oh and by the way, I finally found myself a composer! Turns out one of the guys who composed music for Homestuck liked my game. [URL="https://twitter.com/spellbang"]@spellbang[/URL] is now working on the soundtrack! The song that plays in the video is one of the tracks he made for the game: [URL="https://soundcloud.com/spellmynamewithabang/gore-this-most"]gore this most[/URL]!
I'm pretty happy with the path this game is taking. A demo will come out by july!
[QUOTE=Berkin;51993480]I created my own implementation of Int32.Parse that averages ~700% better performance than the original method. It's missing a few sanity checks, which may slow it down slightly, but it's still quite a significant speedup. Smaller numbers can get as much as a ~2000% speedup.
[URL]https://gist.github.com/TheBerkin/decc733ac59fda3e344a09aa21d599f8[/URL][/QUOTE]
If anyone is curious why he gained such a speed up, you don't get a lot of extra features that come along with .NET's implementation and only need to do simple parse cases such as "1234", you can look at the source here [URL="https://referencesource.microsoft.com/#mscorlib/system/int32.cs,a438016f815a35c3"].NET Reference Source Int32.Parse(String)[/URL].
[QUOTE=Proclivitas;51994701]If anyone is curious why he gained such a speed up, you don't get a lot of extra features that come along with .NET's implementation and only need to do simple parse cases such as "1234", you can look at the source here [URL="https://referencesource.microsoft.com/#mscorlib/system/int32.cs,a438016f815a35c3"].NET Reference Source Int32.Parse(String)[/URL].[/QUOTE]
I was more or less aware of how much stuff is going on behind the scenes but it always surprises me again when I see the code :P
I often deal with serialized numbers that are in a very simple, predictable, format (and if they're not, the whole file is useless anyway) so having a method like Berkin made in the BCL would be cool.
Maybe int.ParseSimple, or as part of some other "helper" class...
Anyway, maybe someone should make a nuget library with super fast methods to parse numbers, do IndexOf, Split, etc. on strings and so on. :P
I've got my slicer finally working with the most complex of shapes. For a long time, it would just quit and start doing really weird things with complex shapes. I couldn't figure out what was wrong, so I rewrote the Polygon and Polygons classes in my application since I thought things were going wrong there, somehow.
[t]http://i.imgur.com/vk1OefA.png[/t]
[t]http://i.imgur.com/EbRYmBY.png[/t]
The simplification is especially great here, as the Douglas-Puecker simplification left 5x as many points for the text, and nearly 7x as many for the other complex object (which is part of a marble track)!
[editline]edit[/editline]
Here's the supremely unfun part: a lot of 3D printers can't print arcs. So every single curve is just a bunch of points, which is really gross and is why 3D printers can be so noisy, especially on detailed curving prints. But some printers can print arcs, and this command works so much better than a series of points that I really need to find some way to identify an arc and parameterize it into the correct curve/arc command to write to the GCode file.
this also mixes with the STL file format being gross. One of the big advantages of AMF is that it defines curves really well in the XML - you can either retrieve a ton of points for a curve, or the data needed to issue an arc command to appropriate 3D printers. But AMF usage is low and not a lot of 3D Modelling/CAD programs support the full featureset offered by AMF.
I'm doing some Pega programming, which is fancy business process modeling. Got an error that said I have -1 errors in 0 places. I guess that's a good thing?
i wanted to learn rust so i made a discord bot that rolls dice and has different configs for different systems
[t]https://dl.dropboxusercontent.com/u/5168294/screencaps/Screenshot%202017-03-22%2013.30.23.png[/t]
fun stuff. rust is really interesting
After four years, Rant can now print highly informational stack traces for runtime errors!
[img]http://i.imgur.com/hficZwe.png[/img]
Officially my new favorite enum name
[t]http://i.imgur.com/pS6f3nP.png[/t]
I giggle like a child every time I see it.
Still tearing my hair out trying to fucking use dev-c++ for the programming class I'm basically the TA for now. Especially because half of the class had problems running the example program the lecturer gave because in the example they had examplestream.open(filepath); which, at least in VS2015 and above where I've tested it, you can do that just fine, open() can be done with just your average std::string. Dev-c++ on the other hand, examplestring.open(filepath.c_str());. And of course this throws compile errors that all these brand ass new to coding students probably couldn't figure out with a decent error log, much less this fucking horrid one.
Worst part is when the programs are graded, they have to open and compile in dev-c++ which means I can't even just tell them to install code::blocks and not suffer in their learning.
Like, it's already a slog to debug your code as a beginner and these poor guys are so fucked because they're having to learn to debug not only their own errors but stupid inane bullshit from the IDE that literally no one can google because the only solid answer to "I have a problem in dev-c++" is, "stop using dev-c++".
Hey guys, my color picker for Unity I made ages ago is now on GitHub and works for 5.5.2
[url]https://github.com/mathiassiig/ColorPickerUnity[/url]
I'll get around to working on it again eventually but it has the pretty basic features like HSL/RGB, alpha and hex.
[QUOTE=F.X Clampazzo;51997485]Still tearing my hair out trying to fucking use dev-c++ for the programming class I'm basically the TA for now. Especially because half of the class had problems running the example program the lecturer gave because in the example they had examplestream.open(filepath); which, at least in VS2015 and above where I've tested it, you can do that just fine, open() can be done with just your average std::string. Dev-c++ on the other hand, examplestring.open(filepath.c_str());. And of course this throws compile errors that all these brand ass new to coding students probably couldn't figure out with a decent error log, much less this fucking horrid one.
Worst part is when the programs are graded, they have to open and compile in dev-c++ which means I can't even just tell them to install code::blocks and not suffer in their learning.
Like, it's already a slog to debug your code as a beginner and these poor guys are so fucked because they're having to learn to debug not only their own errors but stupid inane bullshit from the IDE that literally no one can google because the only solid answer to "I have a problem in dev-c++" is, "stop using dev-c++".[/QUOTE]
This would probably be beyond the ability of the class but I had classes in Uni where they say, "pick any language you want as long as it compiles on linux and uses standard IO", making it independent of everything. That'd require knowledge of compiling tho...
[QUOTE=DoctorSalt;51997670]This would probably be beyond the ability of the class but I had classes in Uni where they say, "pick any language you want as long as it compiles on linux and uses standard IO", making it independent of everything. That'd require knowledge of compiling tho...[/QUOTE]
Yeah, it's just a class on basic c++ that bridges the gap between not taking a coding class at all or having to take a full comp sci class because lots of students in my department are actually process control or other solid 2-year only degrees for industry. So it's there to teach them the basics in case they need it in their jobs.
That's not a good fucking excuse to use fucking dev-c++ though. Like I'm pretty sure the largest and most well funded college in the state could afford the meager cost of education licenses for visual studio. And that's not even considering all the free alternatives that are not fucking trash like dev-c++.
I'm only taking it because this year of classes keeps my loans deferred during my uni switch, but jesus christ I can't imagine coming into this department completely new to it all like these guys are.
Is there any cost to for Visual studio in an educational setting? I thought they were extremely friendly about that (Community edition is free, Dreakspark seems to offer many free things to students/colleges).
[QUOTE=DoctorSalt;51997913]Is there any cost to for Visual studio in an educational setting? I thought they were extremely friendly about that (Community edition is free, Dreakspark seems to offer many free things to students/colleges).[/QUOTE]
There's def. educational licenses for VS, I used to use that before Community was a thing.
i guess i've decided on a name for the forums the player will be visiting in my game:
[t]http://i.imgur.com/YPjGOe0.jpg[/t]
Pufferfish has been heavily inspired by Facepunch! I've even based a few characters on actual FP users. And yes, postal is in it. sorta.
Anyway! more screenshots!
[t]http://i.imgur.com/eqA49dk.jpg[/t]
[t]http://i.imgur.com/4GTvfNk.jpg[/t]
[t]http://i.imgur.com/XzFFW0d.jpg[/t]
[t]http://i.imgur.com/yW31eRL.jpg[/t]
[t]http://i.imgur.com/BDjM1FY.jpg[/t]
[t]http://i.imgur.com/eIhZgbO.jpg[/t]
back to work!
[QUOTE=JWki;51997936]There's def. educational licenses for VS, I used to use that before Community was a thing.[/QUOTE]
The educational licenses cost (the university) money for sure, but the [URL="https://www.visualstudio.com/license-terms/mlt553321/"]Community license[/URL] has an exception that makes it free to use for education.
That still leaves the question whether you want to promote use of a proprietary tool (and the thing still requires online activation, which may be a significant hurdle), but imo that cost is pretty small if you (i.e. the lecturer) otherwise don't have your teaching material in order.
(Imo it's also something you can (should) compromise on in order to provide the best possible education. You can always use fifteen minutes or so to raise awareness of the issue.)
VS being proprietary doesn't matter really imo because if it compiles in VS it probably compiles in something else too. And since everything is basically based on what VS does, you're basically just getting a really good ide that is super user friendly. Especially because most other ides have a conversion for vc project files or you can just set up your own in the new program. But the problem you get is when you're using Dev-c++ or other outdated ides and nothing compiles ever for stupid reasons.
I mean, these guys realistically aren't working in c++98 or 03, or at least they shouldn't be. We're almost to c++17 standard. They should be learning based on compiling for c++14, or 11 at worst, and the rest should be a lecture about version differences so they're aware in the event they need it. Most of these guys probably won't be doing any coding anyway, so pulling their teeth with labs from material that run off newer standards but trying to make them do it in an ide that isn't compiling that standard or newer is beyond retarded.
So, I have two hands. Two separately placed hands, in two different positions, yet the IK system was set to place the elbow/shoulder to where the old hand was, roughly in the middle of the two hands (though not exactly). This meant the right elbow was always a bit close, and the left a bit far
Surprise! I implemented IK for both arms separately. This seems like a rather boring update for a post, so instead what happened was that this broke everything completey. It appeared that when iking to the real hand positions, everything broke. Oopsie. Turns out NaNs were being introduced (also I hate nans). Why were *nans* getting into the code?
Queue debugging montage combined with guiles theme. Although in this case it was the ffx piano works
As it turns out, none of the swanky massive chains of quaternion multiplication were causing this, but for some reason the underlying data has nans in it. ??? How did those get there! A few guard checks and a todo note to fix this idiocy later, independent IK for each hand works! Yay!
Old:
[IMG]https://i.imgur.com/DMiqWuv.png[/IMG]
New (note increased wrist space):
[IMG]https://i.imgur.com/zF1a7B9.png[/IMG]
Stab effect showing quite different elbow positions. This previously would have had both locked to the same position, which is a big step. It adds a little more motion and slightly less stiffness overall to the character which is exactly what the character needs. Still shit, but you have to win some battles:
[IMG]https://i.imgur.com/khkPkVg.png[/IMG]
Oh and also, due to a completely idiotic mistake I made at the beginning, left/right on the character model are swapped. This means the 'left' elbow IKs to the right hand. I'm a human turd. The hands look a little ryu feet sized there, but they're better normally so i suspect its the high FoV (120 horizontal I'm not a pleb)
Still working through internally exactly what to do with the hand models (Edit: This change makes the hands further away from the cylinders and makes it significantly less visually confusing in first person. Still somewhat cluttered but its downgraded from visual catastrophe to "add to the todo list"). On the plus side I can alpha blend textures on the gpu now (hooray!), which means i might just blend a dirt texture into the (procedurally generated) hand colour texture
[QUOTE=Tamschi;51998539]The educational licenses cost (the university) money for sure, but the [URL="https://www.visualstudio.com/license-terms/mlt553321/"]Community license[/URL] has an exception that makes it free to use for education.
[B]That still leaves the question whether you want to promote use of a proprietary tool[/B] (and the thing still requires online activation, which may be a significant hurdle), but imo that cost is pretty small if you (i.e. the lecturer) otherwise don't have your teaching material in order.
(Imo it's also something you can (should) compromise on in order to provide the best possible education. You can always use fifteen minutes or so to raise awareness of the issue.)[/QUOTE]
The real question is why would you care if the tool that you're using is proprietary if it doesn't produce only proprietary content?
Went to a hackathon last weekend, and my team came up with the brilliant idea that people need more Trump in their lives.
Introducing trump text-to-video: [video]https://youtu.be/bevJJ0UGSo4[/video]
Wait did you really sit down and manually make a lookup table of words and their responding clips
[url]https://steamcommunity.com/app/596420/discussions/0/133260492053776048/#c133260492058050009[/url]
Made a suggestion for a fix in Steam Audio and it looks like it'll be fixed for the next release :v:
I really enjoyed being able to spend a work day on this. Work often tends to be the [I]same thing[/I] but now it crossed over into the kind of things I'd actually do back when game development was a hobby.
[QUOTE=NicDasomeone;52000201]Went to a hackathon last weekend, and my team came up with the brilliant idea that people need more Trump in their lives.
Introducing trump text-to-video: [video]https://youtu.be/bevJJ0UGSo4[/video][/QUOTE]
Please release this
[QUOTE=NicDasomeone;52000201]Went to a hackathon last weekend, and my team came up with the brilliant idea that people need more Trump in their lives.
Introducing trump text-to-video: [video]https://youtu.be/bevJJ0UGSo4[/video][/QUOTE]
now do it with phonetic sounds
Sorry, you need to Log In to post a reply to this thread.
What are you working on? v67 - March 2017
