If you have any doubts that the face synthesis one is faked (faked fake?), watch the face of the woman in the bottom left as it loops.
Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
121–129 of 129 posts
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#122If you have any doubts that the face synthesis one is faked (faked fake?), watch the face of the woman in the bottom left as it loops. https://github.com/NVIDIA/vid2vid/blob/master/imgs/face.gif
I'm not following - can you explain?
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#123If you have any doubts that the face synthesis one is faked (faked fake?), watch the face of the woman in the bottom left as it loops. https://github.com/NVIDIA/vid2vid/blob/master/imgs/face.gif
I'm not following - can you explain?
Her(?) face morphs slightly in the first few frames.
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#124What media would someone collect now to be used in the future to reproduce the likeness of loved ones? Video clips of them moving? Talking? Different poses of pictures? Reading the dictionary out loud to get vocal patterns? Heck with impersonating the POTUS. What about a lost friend, sibling or parent?
Black Mirror did this I think - if we can make video, then why not VR (down the line if processing catches up), Second Life iterated. Heaven on Earth?
Also see the film Marjorie Prime.
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#125Earlier quoted context omitted.
There is no competitor, proprietary or open, that comes close to Excel. It's been relentlessly, extensively polished for years and years, and keeps gaining new features every year. And this sticking to the spreadsheet concept, which is very limiting. --- Contrast for example Tableau -- it's a great idea and generated a lot of enthusiasm for a while, but never quite took off as an office package one needs to have. The…
Tableau is great, but it has a much narrower use case: given one or more tables of data, generate graphs for presentation or for exploring the dataset. In comparison, Excel can do that too (just worse), but it can also solve equations, do your company's bookkeeping, and pretty much every other task that relies mostly on numbers. I would argue Open/LibreOffice Calc comes fairly close to Excel if you ignore the worse u…
Yes, but given switching costs and habit formation, why would people care about something that's not strictly Pareto dominant?
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#126Seeing the example of one facial pose video transcribed to three different looking women, I'm imagining a future where Netflix does a/b testing on its shows, using similar tech to swap out different "actors" to find which one resonates with audiences best. They could even generate a new "cast" for each market, after only shooting the show once.
Porn industry will benefit the most.
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#127Earlier quoted context omitted.
But.. if the camera app automatically signs pics/vids, that would mean that a private key is available to the app without any passphrase (or one embedded in the app :O).. So why not just extract they key and sign your fake vid?
Well, for something like this the user's signing key itself would probably be managed by the OS, so not extractable. When the app is done editing, it asks the user via the OS to approve the result and the OS performs the signature. You could even embed a downsample of the original that came signed from the camera hardware.
If the key is managed by the OS but is inaccessible to the user, the concept would seem to be incompatible with free software operating systems. Also, if the camera app can have anything it "makes" approved, the app itself could take a deepfake video (from the web or device storage) and have the OS sign it.
The only way I see it working is if the key is "burnt-in" to in the camera hardware and any applications cannot MitM it.
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#128Earlier quoted context omitted.
Well, for something like this the user's signing key itself would probably be managed by the OS, so not extractable. When the app is done editing, it asks the user via the OS to approve the result and the OS performs the signature. You could even embed a downsample of the original that came signed from the camera hardware.
If the key is managed by the OS but is inaccessible to the user, the concept would seem to be incompatible with free software operating systems. Also, if the camera app can have anything it "makes" approved, the app itself could take a deepfake video (from the web or device storage) and have the OS sign it. The only way I see it working is if the key is "burnt-in" to in the camera hardware and any applications cannot…
Why would you assume it would be inaccessible to the user? I'd expect a key management interface at least. I guess I did say "not extractable", but I meant more that random apps don't have access to it directly, but they call an api to do the signing.
> the app itself could take a deepfake video and have the OS sign it
Note how I said "it asks the user via the OS to approve the result". I would expect a modal OS dialog to let the user review and approve the content before being signed and passed back to the app.
Thinking about it, there's actually nothing stopping this from happening on today's hardware using just application sandboxing. Substitute "OS" above with "Signing App" that does the same thing (accepts media signature requests from other apps, and opens dialog to request approval from user with a preview).
Re: Nvidia Vid2vid: High-resolution photorealistic video-to-video translation
#129Earlier quoted context omitted.
Adobe have definitely bested most oss competitors in their space. Although with the amount of man power at their disposal it would be hard to beat.
The vast bulk of Adobe's advantage is in UX, not technical algorithms. Which makes perfect sense because that tends to be the case with most F/OSS software—technically brilliant but with an face only a programmer could love. Yes, Adobe do have some remarkable algorithms that would be difficult to replicate (e.g. heal brush and content aware fill) but these are a small minority of Adobe's software advantage. The one t…
You're right that UX is one of the biggest problems they have. One thing that is also hard to replicate it how well Adobe's software works together. Embedding smart objects and illustrator files in photoshop documents, right clicking a clip in Premiere and sending it to After Affects and back again without rendering an intermediate file etc.
I would be interested in a Lightroom alternative if anyone can recommend one though.