next »

[Torchkas]

grid breaking pixel art is literally the devil

I mean with camera scrolling and sprite movement.

[Schoq]

me too


Has anyone tried rendering a video with that thing?

[Fallsburg]

It takes 40mn on a i7 with a 970 gpu

FYI, it seems like you aren't using the gpu.  My laptop using the GPU takes closer to 5 minutes, but CPU takes 50ish. Do you have CUDA (or OpenCL) installed properly?  Do you have Theano using CUDA (or OpenCL)?  As a note, I couldn't get Theano to use CUDA (didn't try OpenCL) on windows and had to do it on Linux (Mac would also work).

[Torchkas]

I'm not totally sure how I'd do it in HD anyway. What I'm working on right now is SNES so that will have to use low res scrolling either way.

The fact that scrolling was in HD didn't bother me in FEZ for example. What did was how it used HD rendering for effects and rotating models. You could get rid of both by just changing the resolution it renders in.

[gimymblert]

It takes 40mn on a i7 with a 970 gpu

FYI, it seems like you aren't using the gpu.  My laptop using the GPU takes closer to 5 minutes, but CPU takes 50ish. Do you have CUDA (or OpenCL) installed properly?  Do you have Theano using CUDA (or OpenCL)?  As a note, I couldn't get Theano to use CUDA (didn't try OpenCL) on windows and had to do it on Linux (Mac would also work).

It's the number given by the bot make, might have mis remembered and confuse gpu number for cpu. Haven't made it run yet, mostly because I'm not yet up to date with python and the ton of libraries and installation to do. I'm stuck on trying to have pygame on pycharm, haven't test iddle, and are at lost with all those py file and where to install/put them.

Also marveling at how straightforward vanilla python made it to manipulate text and data (mostly because I got the right tutorial?)

[gimymblert]

https://github.com/jcjohnson/neural-style

On a GTX Titan X, running 1000 iterations of gradient descent with -image_size=512 takes about 2 minutes. In CPU mode on an Intel Core i7-4790k, running the same takes around 40 minutes. Most of the examples shown here were run for 2000 iterations, but with a bit of parameter tuning most images will give good results within 1000 iterations.

 

[Fallsburg]

As a note, the one you just listed uses Torch (a variant of Lua) (although it awkwardly has to use a model from Caffe [another neural network framework]) instead of Python. 

I've been more a fan of Torch than Theano (or the other Python frameworks) for Neural Net stuff, but Python is just a much more robust programming environment. 

[gimymblert]

another fangame turn its own franchise with the blessing of the creator of huge franchise who actually did help in the design

[Torchkas]

That art is really well done.

[ProgramGamer]

another fangame turn its own franchise with the blessing of the creator of huge franchise who actually did help in the design

Formerly known as "My Little Pony: Fighting Is Magic", this amazing fan game was originally C&D'd by Hasbro. However, they decided to re-skin the project and keep working on it and they, amazingly, managed to hire Lauren Faust to work on the character design for their game. And now they have a crowdfunding campaign starting in 13 days.

Just in case you wanted more info.

[s0]

As a note, the one you just listed uses Torch (a variant of Lua) (although it awkwardly has to use a model from Caffe [another neural network framework]) instead of Python. 

I've been more a fan of Torch than Theano (or the other Python frameworks) for Neural Net stuff, but Python is just a much more robust programming environment. 

why isnt there just a compiled program that u can dl

i have no intention of ever learning python i just wanna mix some pix together

[Fallsburg]

You can tweet at DeepForger with a source image and style image and it will combine them, but that's about as plug and play as you can get right now.

[gimymblert]

As a note, the one you just listed uses Torch (a variant of Lua) (although it awkwardly has to use a model from Caffe [another neural network framework]) instead of Python. 

I've been more a fan of Torch than Theano (or the other Python frameworks) for Neural Net stuff, but Python is just a much more robust programming environment. 

why isnt there just a compiled program that u can dl

i have no intention of ever learning python i just wanna mix some pix together

I'm investigating the cause, I'm not sure you need to learn the entirety of python, things do got dispersed a bit and thing don't exactly work out of the box (yet in my case)

[mks]

They started selling christmas sweets at my local supermarket. What a crazy world...

[s0]

https://www.psychologytoday.com/blog/give-and-take/201309/goodbye-mbti-the-fad-won-t-die

just leaving this here

[gimymblert]

rip MBTI welcome big5

[Cobralad]

well well well
http://www.allgamesbeta.com/2015/09/bandai-namco-trademarks-burning-blood.html

[gimymblert]

what are we supposed to see?

[Pfotegeist]

Begins monologue: I'm better than a year ago. I can tell because I don't feel like my chair is trying to break my back. Eating green leafy veggies helped. I can tell because I rarely ate any until a year ago. I didn't need to eliminate anything out of my diet except what I'm allergic to.  The veggies ended up suppressing hunger pains which indicated malnutrition. Also I like sprouted grains, but I can't guarantee they matter, the public awareness (about sprouting) is being raised quickly enough. There's still very little advertisement or product featuring green veggies. End of monologue.

[Cobralad]

what are we supposed to see?

bamco namdai trademarks "slashy souls"