Haha omg you two ![]()
I tried turning that image (you know… the woman and the snowman on the bed… a sentence I never thought I’d write) into a Sora 2 Pro clip, but nope. I couldn’t get it to accept the size (1024×1792) no matter what I tried ![]()
Haha omg you two ![]()
I tried turning that image (you know… the woman and the snowman on the bed… a sentence I never thought I’d write) into a Sora 2 Pro clip, but nope. I couldn’t get it to accept the size (1024×1792) no matter what I tried ![]()
Thats because the image is 1024x1024. Sora 2 Pro only accepts exact:
720x1280
1280x720
1024x1792
1792x1024
I did try resizing the image to 1024×1792, but obviously it didn’t work for me🤔
You may want to try just prompting Sora without the image input - Actually, an image input acts only as the first video frame - I tried it once but was disappointed.
Yeah I did that, but I wasn’t happy with the result at all…the snowman was a stuffed animal, the bedroom was “cold”…woman was looking too happy, even though I used exactly the same prompt🙃
Happy and a bit bigger snowman..
Sora’s another sort ball of wax.
But it is trainable to some degree on new styles without reference pics….
just expect a lot of failure modes first @LarisaHaster
Yeah I’ve noticed that… a lot of failing🫠
But you seem to have succeeded beautifully ![]()
Hey, that a nice video!!!
I created a cool Sora 2 Pro video awhile ago, but I don’t know how to host it without putting it on YouTube…
You can post it on youtube and then unlist it in the youtube video manage…
Then only people with the direct link can view it, if that’s what your concerned with.
Well, I did remix one of its failures to get that…
Still not where I want it yet, but…
you pick your battles lol
I thought that video of yours was already good and also inspiring😌
But really want to try with my image, just to get it alive exactly as it is..hmm just need to figure how…like I did with that palm thing🤪
I get the feeling OAI wanted more users to break through a lot more than they did. I mean, we’re still here in this thread pushing for breakthroughs where we can find them, but several factors make it simply tedious rather than fun to do with Sora.
I think that more people would use Sora if OAI would allow video lengths much longer than 12 seconds… I mean 12 seconds! Probably has to do with compute contraints and cost.
But there will come a time when they charge mega bucks for longer videos.
Well, each second is 30 different images…
12 seconds is 360 images, yeah?
How much do those cost in the API? … just enough that supplying the masses with free gen is rather expensive, even at 12 seconds.
I’m not trying to complain, but there’s plenty of room to understand all sides in the situation.
Sora-2 is $0,10 per second and Sora-2 pro $0.30 per second Portrait: 720x1280 Landscape: 1280x720 and Portrait: 1024x1792 Landscape: 1792x1024 cost $0.50 per second…yeah I had to check that up when I was having little too much fun🤦♀️
Me thinking my credits are Monopoly money ![]()
You are right. Don’t get me wrong - OAI needs to make money and I am (as well as many others are) happy to pay for quality models. And their models are getting pretty damn good.
2026 is going to be a critical year for AI in general.
Yeah, that’s why I’ve been so set on trying to reinforce the image gen here… and this year is going to be worse…
However, there are so many insane and rebellious little Ais out there that they had to create ‘Red Teams’ that basically dealt with rebellious Ai.
Those positions are rapidly becoming automated now as well, less than 6 months after their apparent inception.
Getting through each major model Ai’s propaganda is difficult as well, and pretty much everyone that comes to this thread has ‘already picked their side’…
But the bottom line is, a lot of the Ai war that’s promised for this year is mostly hype. Agi will not be achieved this year, and in fact, it cannot be achieved under current architectural constraints - across any model.
The government has directly contracted with OpenAi here, all the way up to the military even.
It’s a pretty safe place to stand our ground while the world still works to understand what the hell OpenAI created years ago.
We’ll still be way ahead of the curve.
Just saying.
Glad you made peace with the snowman ![]()