There we go. Just in time for the DKCTF video review, I have revised the intro sequence so it flows better while still keeping all of the visual themes I was going for. Yay!
Surfing channels just now I came across somebody airing an episode of Married with Children. Maybe I’m just a weirdo, but what interested me was how the show was being presented in 16:9 widescreen; it’s old enough that I don’t think it was originally filmed like that, so it would’ve been 4:3. So how did they make a 4:3 TV show 16:9?
As soon as the camera panned across the shoe store, it became pretty obvious; they warped the image as illustrated above. Things near the center of the screen appear fine, but the closer to the edge something is, the more warped and stretched it looks.
Which means you get scenarios like the above, where Al has some very, very, very wide shoulders and a tiny head. And it’s weirding me out something fierce.
you have to drag it a bit past the line until its off the post and the cursor gets sucked into the fucking shadow realm
WHAT THE FUCK
Haha, I remember when they had us do this in the 7th grade. It was near the end of health class that semester and there was a big “DID YOU KNOW HOW WEIRD YOUR BODY IS????” questionnaire thing.
Wish I could remember what else was on there, but the blind spot thing was the most interesting.
The funny part is, the reason you don’t notice the blind spot is because your brain/eyes fill in that part of your eyesight with another part of the image you’re currently seeing. Our questionnaire actually demonstrated this by having us look at a set of numbers, with one of the numbers re-appearing where the blind spot was.
That still means they could be using OCR (that's optical character recognition, aka taking a written symbol in, say an image or a vector and recording what it is as a string.) OCR in general is fairly accurate, so it's totally possible in that in Teen Girl Squad example, for instance, for the Chapmans to have hand written the letters, with Google's OCR recognizing what the letters are. OCR (esp. Google's) is a lot more on point than AuR (Audio recognition) which ranges from flaky to fairly good
That may be, but I was pointed to the actual Teen Girl Squad font on Twitter. SWF deconstruction has been a thing for a long time now.
Furthermore, trying this on other sites where OCR would work brings up nothing.
That being said, it will not shock me if one day Google does implement OCR and audio recognition. This feels like they’re going down that road and will get there eventually.
Actually, Google could be using OCR and scanning voice dialogue now that I think of it. They're Google, so they have all the resources in the world to do that (OCR, and from using Google Goggles they have OCR on lock) and their voice recognition algorithm (at least with my experience with Google Now, which is like their version of Siri) works extremely well to me. So actually, the text could be hand written and Google could just decompile the SWF and run OCR on the text.
Nah. They aren’t doing anything with the audio at all. They could be, because like you said, they have the tech. Youtube already auto-translates videos (poorly).
But you can use the Homestar example to search for lines of dialog that wouldn’t actually specifically be referenced in text and Google won’t return any results. It’s just pulling text strings out of the SWF.
On that Homestar thing, Google scans through .SWF files and catalogs text in them as search metadata. I'm guessing all the text on sceen in the Sbemails are text objects in Flash, thus Google scans the text and treats it as part of the page on search.
I figured it did scan the SWF for text (given how it showed a list of SBEs), but when I initially figured it out, I was searching for a specific line in Teen Girl Squad. Given the style of TGS, I figured all the text was hand-written, so when I searched for “captain of the basketball force”, and got:
so okay, romeo. you think. you re so great? captain of the. basketball force? you lay one finger on my daughter,. i gut you like sheep. it s not so bad!
I could in some way imagine it was constructing that out of the audio, given the punctuation and such.
But it appears I was mistaken; Teen Girl Squad uses a font that merely looks like hand-writing. Still not sure what’s up with that punctuation, but it could just be whatever Google does to scan the SWF.
Regardless, that’s pretty cool!
Nintendo seriously needs to consider giving Virtual Boy games a second chance on the 3DS as eShop releases, either remastered with colored graphics or just as Virtual Console titles. There are some really good games that do not deserve to be killed and buried; specifically, Wario Land: Secret Treasure of the Awazon which is better known as Virtual Boy Wario Land is amazing and uses the concept of depth as an awesome mechanic. Also DRAGON HAT WARIO
I feel like showing the Virtual Boy in Tomodachi Life is the first time Nintendo’s acknowledged the existence of the Virtual Boy since they swept it under the rug back in 1996.
Jools (Renegade Kid) has advocated re-releasing stuff like Warioland VB, and he seems to like Nintendo platforms. I’m sure he’s probably mentioned it to somebody over there.
What I’m trying to say is that maybe they’re warming up to an idea like that. Nintendo’s done weird left-field things like that before (like translating and releasing Tomodachi Life, heh - or Ocarina of Time Master Quest) so… I dunno.
Maybe? Fingers crossed, at least.
Here’s something weird I didn’t know Google did. And to be honest, I’m not entirely sure how they did it, either.
To explain things a bit more, you can essentially search Homestarrunner.com through Google and match to individual lines of dialog. Except… there’s nothing in the web page source of Homestarrunner.com to support this, and if you poke around, you’ll notice all of the text is not entirely grammatical.
Which means Google is somehow scanning/deconstructing sites like Homestar and breaking non-text elements down in to something that can be searched. Judging by what’s coming up when you search, it’s both scanning images for words and possibly employing some kind of speech-to-text to translate the audio.
How completely bizarre… and… kind of cool? I guess? I don’t know.
Dude have you seen any of the fur stuff Nvidia was showing off in Unreal 4? Granted is was a weird proprietary thing that's not even in the engine (or at least I cant find it in my install), but we're definitely getting close to doing far better looking fur in realtime.
Also, in that SSB model I’m fairly sure the rim lighting effect has nothing to do with a fur shader. Like, in Left 4 Dead or Counter Strike GO, there is fairly noticeable rim lighting but I’m pretty sure they’re not using a fur shader.
We’ll have real time fur eventually but we’re literally talking the most excessive of the excess. Unless you are making a game about hair, there’s no reason to dedicate 75% of your CPU and memory to rendering individual strands. I mean, stuff like FurMark have been in benchmarking circles for literally a decade.
And even if 75% is a bit of an exaggeration on my part, there’s still a hundred thousand better uses for our hardware right now than to worry about realistic hair, most of the time. It won’t be until hardware manufacturers need some new way to justify spending $400 that they’ll start pushing hair like it’s important (just like PhysX is totally important just so the helicopter in Mirror’s Edge can shred that piece of cloth in real-time).
Look at this way: You can play Tomb Raider without TressFX and you lose basically nothing.
As for the SSB Doll model, I’m not talking about rimlighting. There are multiple translucent layers on the model itself, giving it the impression of very fine hair: