MATL online working great on this one!

@LuisMendo Ha! that's the type of question where it really will be useful

definitely a leg up on most other languages' interpreters

I want to tweak how the iamges are displayed a little bit. I want them to max out at their native resolution

but for images created with imwrite, i'd like for them to be a little bigger so you can actually see them

And I need it to handle portrait and landscape images.

After that, I think I've added most pressing things at this point. Are there many major issues that you think we should deal with before unleashing this on the world?

I was also going to write some unittests

@Suever Yes, images with imwrite tend to be quite small (usually a few tens of pixels wide/high)

@Suever I don't think there are any issues

Yea and the way I have it now, I just scale everything to 75% of the output width

which leads to some strange things: here

@Suever I see. One possibility is to leave the decision to the user: a toggle button to activate or deactivate scaling. But I don't know if it's easier to create that button (and make it work) or to find a formula to decide which resolution to use

Hey! Now with favicon! :-)

A matrix. Very nice

Hahaha yea I wanted to have SOMETHING and MATL won't fit on one line cleanly in a 16x16 grid

we can change it if desired, I just wanted a placeholder

Yes, I'd also like it on a single line, but it won't fit :-)

M's are just too broad

Oh I also added the paste input functionality

it's the little + to the bottom right of the input block

I didn't do extensive testing yet but it seemed to work ok

Oh! How nice!

And discreet. You seem to like letting people be surprised as they find out, haha

haha I make the features for myself and if others discover them, they win

I tend to be quite the minimalist so I'd rather keep functionality subtle

Yes, "minimalist" and "subtle" are the words I was looking for :-)

Don't want to end up with an interface like this: createdigitalmusic.com/app/uploads/2016/03/screen.png

Hahaha. BTW, do you use that kind of programs? I use Cubase myself

Also most of the "advanced" features are really only needed by people crafting answers using the site

I have played around with some of them before but not extensively

Do you play some intrsument? I play guitar. And I've also used Matlab to synthesize bass and compose drum parts (which was terribly funny!)

Yea actually. I played piano when I was a kid and have been off and on since. Then picked up drums and all other percussion when I was 12 or so. Haven't had a chance to play much lately sadly.

Haha never thought to use MATLAB as a synthesizer but really it's setup quite well to do that

It was more of a sampling thing: take some waveform and then change its envelope or pitch to produce all required notes

An example is the bass part here (not very loud, you may have to struggle to hear it)

hahah that's awesome

Have you ported it to MATL yet?

I would love to learn to play guitar. Sadly I have very stubby fingers which make getting around the fret board a little tricky

@Suever That would really be an interesting next level, hahaha

ooooo we should make a challenge that involves making sounds

or output a sound file or whatever

probably rules out a lot of languages sadly :(

I thought of including that in MATL, but I ended up discarding it. Well, I guess, introducing audioread and audiowrite would suffice. The bad part is that the results can't be easily posted (and of course not tested online)

Well until there was my interpreter :)

I thought there were some imgur sites for sound clips

Huh? Audio?

Oh!

The interpreter could definitely support audio if the language did

not sure if it's worth it though

unless we have a clipboard for "anonymous functions"

I am already envisioning some FM synthesis challenge. That's easy to do and gives good results with little complexity

Nah, probably not worth it

@Suever ?

Oh just like we have pre-defined literals, you'd have a "register" of functions

1Y9 could be audioread

2Y9 could be audiowrite

because that's not confusing or anything

haha

I still think that it would be possible to have "modes" similar to vim

Ah, well, that's more or less what YG and ZG do

shift into and out of modes

image processing mode, string mode, date mode, audio mode

We could have one two-char statment for different audio functions. But before that we've yet to decide on some image processing functions

Oh yea just like YG and ZG, just more functions per

Yep

yea I need to spend some time thinking about the image processing ones

Anyway, for audio I don't see many functions required

there were really one 2-3 that I thought could be useful

Doesn't look like we have fft

maybe that wouldn't be that useful, I'm not sure

@Suever and maybe ask Ray. That would be a good excuse to get him involved in MATL :-) He was in the beginning but not so much lately (well, as with SO)

I should log what is searched for in the search box so we can remember what we were hoping existed that didn't

@Suever Oh, that's on my to-do list actually

Nice thing is we can skip ifft

Yea I think I've looked for that a few times

(duality)

Would the default output have fftshift applied?

I though of putting fft as a 1$ version of convolution Z+. That would make sense, as FFT and convolution are closely tied. But we need more than 1 input for FFT

@Suever Hm, good point. fftshift may be needed too

Yea I was just thinking that actually

Yea I guess it depends what it's to be used for

non-shifted is nice if you just want the amplitude at a given frequency because it's easy to find

I don't have any good ideas of how it would be used at this time though

We could probably define only their 2D versions. I think the 1D versions are particular cases of the 2D onesm like convolution

I do know I was looking for it when I was writing a classifier for the upgoat/downgoat question

@Suever Haha, you tackled that??

well I wrote one and it did mostly OK. There were some improvements it needed so I had shelved it

I was just using the fft to get spatial frequencies from top and bottom of hte image separately

A few of the test cases really messed it up though

It seemed to me like a very hard challenge

Except for Mathemetica of course

I'm tired of its builtins

hahah yea that blew my mind

The concept of my approach was pretty straightforward and relied on the fact that you would have a higher amplitude of horizontal spatial frequencies where the legs were (due to grass and the legs being vertical)

So I believe I took some ratio of horizontal frequency power from the bottom part of the image to the top part

Pretty naive approach but it did ok. I think like 100% of all the first test cases and 70% of the others

That sounds like a very good approach. Simple and effective

Yea I love fourier transform-based stuff like that. No better way to get global information in a sparse domain

Sometimes I wish people would answer these questions with really golfy answers which clearly aren't what the instructor wanted and it would be clear that the student didn't do it

@Suever Conversion from Python array to MATL(AB)'s only works for two levels, obviously. Should that be indicated? Not sure, perhaps it would add too much clutter. And I don't even know if [[1] [2]] is called "two nesting levels" or "one nesting level" :-D

@LuisMendo Ah yea maybe instead of "Python Array" make it 2D array?

@Suever Maybe... but it would also be important to note the origin (Python format)