Jonathan Stephens
@jonstephens85
Followers
11K
Following
6K
Media
2K
Statuses
10K
Radiance Fields | Computer Vision | GenAI | Chief Evangelist at @EveryPointIO | Host of the Computer Vision Decoded Podcast
Salem, OR
Joined February 2009
A really cool parallax movement affect using a zoom out with @NVIDIAAIDev's GEN3C. All generated from an archival photo. Notice the couple of guys sitting on the Sphinx. Wild!.#GENAI #AI
0
2
6
Generated off of a single image of a @SpaceX launch. I meant to pan the other direction, but this turned out amazing nonetheless!. I am absolutely loving pushing the boundaries of @NVIDIAAIDev's GEN3C. #GENAI #Computervision #VFX
0
4
70
The affect of the movement distance modifier on @NVIDIAAIDev's GEN3C. Default 0.3, 0.6, and 1.2. #Computervision #AI #GenAI
3
7
70
This is one heck of a first tweet!.
🚀 My first tweet!.(1/n) Thrilled to share our new work: Context-as-Memory (CaM) — tackling the memory problem in Video World Model!. Our idea: context=memory. By leveraging context, CaM preserves consistency across generations (like Genie 3). 🎥 Check out our demo video below!
0
0
5
I'm finally getting decent output using GEN3C and dynamic video input. The biggest change I made was use Video Depth Anything to get the temporally consistent depth maps. Interesting that it simulated camera lens flare. The trees also get a bit wavy. #GenAI #Computervision
0
3
82
Splatfacto was the best output so far. I need to mask myself out of every image though as I am messing up the scene!. #computervision
1
0
29
Got COLAMP Native 360 -> Nerfstudio working. I don' think this is the best workflow unfortunately. Unless I can figure out how to mask myself out.
4
8
56
More GEN3C dynamic video generation fails. I am learning that the SLAM model is not as important. DepthAnythingv2 really struggled! I may have some good outputs coming soon.
4
1
56
My first two attempts last night at using dynamic videos with GEN3C. While both are obviously turned out a "failure", I learned how to prep the data and also learned how important having good SLAM data is. I used MiDaS for monocular SLAM and the depth data was not stable. Next
8
6
77
The team at @NVIDIAAIDev really nailed it with ViPE!! This SLAM system generated pose, depth, point clouds, etc, in a few seconds! Now to get the output to work with GEN3C. If you are interested in running this yourself, it took about 5 minutes to get up and running from
11
58
485
RT @huangjh_hjh: [1/N] 🎥 We've made available a powerful spatial AI tool named ViPE: Video Pose Engine, to recover camera motion, intrinsic….
0
98
0
RT @janusch_patas: Evaluating Fisheye-Compatible 3D Gaussian Splatting Methods on Real Images Beyond 180 Degree Field of View. Contribution….
0
13
0
RT @ZGojcic: 3DGRT is now available also with Vulkan API as part of the Vulkan Gaussian Splatting: .
github.com
Sample implementing several rendering methods for 3D gaussians using Vulkan API - nvpro-samples/vk_gaussian_splatting
0
7
0
On a scale of 1-10, 10 being best, how well did @NVIDIAAIDev's GEN3C handle that large mirror reflection? I think it did a great job! Only one photo was used for input. #3D #Computervision #GENAI
7
4
103
More tests last night with @NVIDIAAIDev's GEN3C. This was all generated from a single photo using local compute. I love how it accurately simulated the light fade around the building corner. Next up, dynamic scenes. #3D #GenAI #Computervision
25
141
1K
Serious question, have any of you got ChatGPT-5 yet? My pro plan hasn't updated yet ðŸ˜.
7
0
1
Fitting. My wife is the 15,000th follower. She doesn’t watch anything I make on any platform 🤣
0
0
5
