Can Video Diffusion Model Reconstruct 4D Geometry?
Jinjie Mai, Wenxuan Zhu, Haozhe Liu, Bing Li, Cheng Zheng, Jürgen Schmidhuber, Bernard Ghanem
tl;dr: pretrained video VAE->finetune pointmap VAE; finetune Open-Sora within latent space of video&pointmap
arxiv.org/abs/2503.21082
Jinjie Mai, Wenxuan Zhu, Haozhe Liu, Bing Li, Cheng Zheng, Jürgen Schmidhuber, Bernard Ghanem
tl;dr: pretrained video VAE->finetune pointmap VAE; finetune Open-Sora within latent space of video&pointmap
arxiv.org/abs/2503.21082
March 28, 2025 at 7:30 AM
Can Video Diffusion Model Reconstruct 4D Geometry?
Jinjie Mai, Wenxuan Zhu, Haozhe Liu, Bing Li, Cheng Zheng, Jürgen Schmidhuber, Bernard Ghanem
tl;dr: pretrained video VAE->finetune pointmap VAE; finetune Open-Sora within latent space of video&pointmap
arxiv.org/abs/2503.21082
Jinjie Mai, Wenxuan Zhu, Haozhe Liu, Bing Li, Cheng Zheng, Jürgen Schmidhuber, Bernard Ghanem
tl;dr: pretrained video VAE->finetune pointmap VAE; finetune Open-Sora within latent space of video&pointmap
arxiv.org/abs/2503.21082
VGGT: Visual Geometry Grounded Transformer
@jianyuanwang.bsky.social Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, David Novotny
tl;dr: Mast3r meets CoTracker, overcomplete predictions (depth AND pointmap, etc).Training: 64A100x9days. Sota on IMC-2020.
arxiv.org/abs/2503.11651
@jianyuanwang.bsky.social Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, David Novotny
tl;dr: Mast3r meets CoTracker, overcomplete predictions (depth AND pointmap, etc).Training: 64A100x9days. Sota on IMC-2020.
arxiv.org/abs/2503.11651
March 17, 2025 at 3:27 PM
VGGT: Visual Geometry Grounded Transformer
@jianyuanwang.bsky.social Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, David Novotny
tl;dr: Mast3r meets CoTracker, overcomplete predictions (depth AND pointmap, etc).Training: 64A100x9days. Sota on IMC-2020.
arxiv.org/abs/2503.11651
@jianyuanwang.bsky.social Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rupprecht, David Novotny
tl;dr: Mast3r meets CoTracker, overcomplete predictions (depth AND pointmap, etc).Training: 64A100x9days. Sota on IMC-2020.
arxiv.org/abs/2503.11651
G3T Up! Gravity Aligned Coordinate Frames Simplify Pointmap Processing
Bharath Raj Nagoor Kani, @snavely.bsky.social
tl;dr: fine-tune VGGT on custom gravity-aligned 3D data->gravity-aligned pointmaps
arxiv.org/abs/2605.27372
Bharath Raj Nagoor Kani, @snavely.bsky.social
tl;dr: fine-tune VGGT on custom gravity-aligned 3D data->gravity-aligned pointmaps
arxiv.org/abs/2605.27372
May 27, 2026 at 1:07 PM
G3T Up! Gravity Aligned Coordinate Frames Simplify Pointmap Processing
Bharath Raj Nagoor Kani, @snavely.bsky.social
tl;dr: fine-tune VGGT on custom gravity-aligned 3D data->gravity-aligned pointmaps
arxiv.org/abs/2605.27372
Bharath Raj Nagoor Kani, @snavely.bsky.social
tl;dr: fine-tune VGGT on custom gravity-aligned 3D data->gravity-aligned pointmaps
arxiv.org/abs/2605.27372
C3Po: Cross-View Cross-Modality Correspondence by Pointmap Prediction
Kuan Wei Huang, Brandon Li, Bharath Hariharan, @snavely.bsky.social
tl;dr: floorplan (very paper-ish style) to 3D model dataset + model (finetuned DUSt3R) trained on it.
Cool stuff.
arxiv.org/abs/2511.18559
Kuan Wei Huang, Brandon Li, Bharath Hariharan, @snavely.bsky.social
tl;dr: floorplan (very paper-ish style) to 3D model dataset + model (finetuned DUSt3R) trained on it.
Cool stuff.
arxiv.org/abs/2511.18559
November 26, 2025 at 8:42 PM
C3Po: Cross-View Cross-Modality Correspondence by Pointmap Prediction
Kuan Wei Huang, Brandon Li, Bharath Hariharan, @snavely.bsky.social
tl;dr: floorplan (very paper-ish style) to 3D model dataset + model (finetuned DUSt3R) trained on it.
Cool stuff.
arxiv.org/abs/2511.18559
Kuan Wei Huang, Brandon Li, Bharath Hariharan, @snavely.bsky.social
tl;dr: floorplan (very paper-ish style) to 3D model dataset + model (finetuned DUSt3R) trained on it.
Cool stuff.
arxiv.org/abs/2511.18559
C3Po: Cross-View Cross-Modality Correspondence by Pointmap Prediction
Kuan Wei Huang, Brandon Li, @bharathhariharan.bsky.social, @snavely.bsky.social
tl;dr: in title; paired floor plans and ground-view photos with annotated correspondences and poses
arxiv.org/abs/2511.18559
Kuan Wei Huang, Brandon Li, @bharathhariharan.bsky.social, @snavely.bsky.social
tl;dr: in title; paired floor plans and ground-view photos with annotated correspondences and poses
arxiv.org/abs/2511.18559
November 25, 2025 at 3:09 PM
C3Po: Cross-View Cross-Modality Correspondence by Pointmap Prediction
Kuan Wei Huang, Brandon Li, @bharathhariharan.bsky.social, @snavely.bsky.social
tl;dr: in title; paired floor plans and ground-view photos with annotated correspondences and poses
arxiv.org/abs/2511.18559
Kuan Wei Huang, Brandon Li, @bharathhariharan.bsky.social, @snavely.bsky.social
tl;dr: in title; paired floor plans and ground-view photos with annotated correspondences and poses
arxiv.org/abs/2511.18559
ahh this is my mistake - it is just image, pointmap datapoints
I am still impressed by sora’s ability to learn 3D without 3d info (like pointmaps)!
I am still impressed by sora’s ability to learn 3D without 3d info (like pointmaps)!
March 11, 2025 at 2:10 PM
ahh this is my mistake - it is just image, pointmap datapoints
I am still impressed by sora’s ability to learn 3D without 3d info (like pointmaps)!
I am still impressed by sora’s ability to learn 3D without 3d info (like pointmaps)!
tl;dr: panorama->four perspective views->virtual camera rig->local pose/pointmap repair->loop retrieval->Sim(3) graph optimization
September 16, 2026 at 3:36 PM
tl;dr: panorama->four perspective views->virtual camera rig->local pose/pointmap repair->loop retrieval->Sim(3) graph optimization
POMATO: Marrying Pointmap Matching with Temporal Motion for Dynamic 3D Reconstruction
Songyan Zhang, Yongtao Ge, Jinyuan Tian, Guangkai Xu, Hao Chen, Chen Lv, Chunhua Shen
tl;dr: DUSt3R+matching head->pairwise input; DUSt3R+temporal head->sequential input
arxiv.org/abs/2504.05692
Songyan Zhang, Yongtao Ge, Jinyuan Tian, Guangkai Xu, Hao Chen, Chen Lv, Chunhua Shen
tl;dr: DUSt3R+matching head->pairwise input; DUSt3R+temporal head->sequential input
arxiv.org/abs/2504.05692
April 9, 2025 at 3:36 AM
POMATO: Marrying Pointmap Matching with Temporal Motion for Dynamic 3D Reconstruction
Songyan Zhang, Yongtao Ge, Jinyuan Tian, Guangkai Xu, Hao Chen, Chen Lv, Chunhua Shen
tl;dr: DUSt3R+matching head->pairwise input; DUSt3R+temporal head->sequential input
arxiv.org/abs/2504.05692
Songyan Zhang, Yongtao Ge, Jinyuan Tian, Guangkai Xu, Hao Chen, Chen Lv, Chunhua Shen
tl;dr: DUSt3R+matching head->pairwise input; DUSt3R+temporal head->sequential input
arxiv.org/abs/2504.05692
PLANA3R: Zero-shot Metric Planar 3D Reconstruction via Feed-Forward Planar Splatting
Changkun Liu, Bin Tan, Zeran Ke, Shangzhan Zhang, Jiachen Liu, Ming Qian, Nan Xue, Yujun Shen, Tristan Braud
tl;dr: MAST3r, but predicts planar patches instead of pointmap
arxiv.org/abs/2510.18714
Changkun Liu, Bin Tan, Zeran Ke, Shangzhan Zhang, Jiachen Liu, Ming Qian, Nan Xue, Yujun Shen, Tristan Braud
tl;dr: MAST3r, but predicts planar patches instead of pointmap
arxiv.org/abs/2510.18714
November 27, 2025 at 9:41 AM
PLANA3R: Zero-shot Metric Planar 3D Reconstruction via Feed-Forward Planar Splatting
Changkun Liu, Bin Tan, Zeran Ke, Shangzhan Zhang, Jiachen Liu, Ming Qian, Nan Xue, Yujun Shen, Tristan Braud
tl;dr: MAST3r, but predicts planar patches instead of pointmap
arxiv.org/abs/2510.18714
Changkun Liu, Bin Tan, Zeran Ke, Shangzhan Zhang, Jiachen Liu, Ming Qian, Nan Xue, Yujun Shen, Tristan Braud
tl;dr: MAST3r, but predicts planar patches instead of pointmap
arxiv.org/abs/2510.18714
LONG3R: Long Sequence Streaming 3D Reconstruction
Zhuoguang Chen, Minghui Qin, Tianyuan Yuan, Zhe Liu, Hang Zhao
tl;dr: all spatio-temporal 3D memory+current image->memory gating->relevant memory->dual-source refined decoder->pointmap
arxiv.org/abs/2507.18255
Zhuoguang Chen, Minghui Qin, Tianyuan Yuan, Zhe Liu, Hang Zhao
tl;dr: all spatio-temporal 3D memory+current image->memory gating->relevant memory->dual-source refined decoder->pointmap
arxiv.org/abs/2507.18255
July 25, 2025 at 12:45 PM
LONG3R: Long Sequence Streaming 3D Reconstruction
Zhuoguang Chen, Minghui Qin, Tianyuan Yuan, Zhe Liu, Hang Zhao
tl;dr: all spatio-temporal 3D memory+current image->memory gating->relevant memory->dual-source refined decoder->pointmap
arxiv.org/abs/2507.18255
Zhuoguang Chen, Minghui Qin, Tianyuan Yuan, Zhe Liu, Hang Zhao
tl;dr: all spatio-temporal 3D memory+current image->memory gating->relevant memory->dual-source refined decoder->pointmap
arxiv.org/abs/2507.18255
VGGT evals
4) Kyiv puppet theater (IMC23-train) - great quality
5) Charles Square church - camera poses seems to be good, but unprotecting visualization artifacts are annoying. Pointmap prediction is much cleaner (3rd image)
2/
4) Kyiv puppet theater (IMC23-train) - great quality
5) Charles Square church - camera poses seems to be good, but unprotecting visualization artifacts are annoying. Pointmap prediction is much cleaner (3rd image)
2/
March 17, 2025 at 4:21 PM
VGGT evals
4) Kyiv puppet theater (IMC23-train) - great quality
5) Charles Square church - camera poses seems to be good, but unprotecting visualization artifacts are annoying. Pointmap prediction is much cleaner (3rd image)
2/
4) Kyiv puppet theater (IMC23-train) - great quality
5) Charles Square church - camera poses seems to be good, but unprotecting visualization artifacts are annoying. Pointmap prediction is much cleaner (3rd image)
2/
naver has a patent app. on the pointmap approach, let's see what happens I guess?
August 4, 2025 at 12:20 PM
naver has a patent app. on the pointmap approach, let's see what happens I guess?
Thank you! And yes same as pointmap matching (fig2, theta) it’s minimising alpha not beta (since rays are normalised, we can use eq 3)
December 17, 2024 at 10:20 AM
Thank you! And yes same as pointmap matching (fig2, theta) it’s minimising alpha not beta (since rays are normalised, we can use eq 3)
Interesting they still have a "pointmap" loss term though skipping the head. Also glad they got rid of the horrid multi-step pass through the camera head.
May 15, 2026 at 7:01 PM
Interesting they still have a "pointmap" loss term though skipping the head. Also glad they got rid of the horrid multi-step pass through the camera head.
𝗣𝗢𝗠𝗔𝗧𝗢: 𝗠𝗮𝗿𝗿𝘆𝗶𝗻𝗴 𝗣𝗼𝗶𝗻𝘁𝗺𝗮𝗽 𝗠𝗮𝘁𝗰𝗵𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗧𝗲𝗺𝗽𝗼𝗿𝗮𝗹 𝗠𝗼𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝟯𝗗 𝗥𝗲𝗰𝗼𝗻𝘀𝘁𝗿𝘂𝗰𝘁𝗶𝗼𝗻
Songyan Zhang, Yongtao Ge, Jinyuan Tian ... Chunhua Shen
arxiv.org/abs/2504.05692
Trending on www.scholar-inbox.com
Songyan Zhang, Yongtao Ge, Jinyuan Tian ... Chunhua Shen
arxiv.org/abs/2504.05692
Trending on www.scholar-inbox.com
April 10, 2025 at 6:00 AM
𝗣𝗢𝗠𝗔𝗧𝗢: 𝗠𝗮𝗿𝗿𝘆𝗶𝗻𝗴 𝗣𝗼𝗶𝗻𝘁𝗺𝗮𝗽 𝗠𝗮𝘁𝗰𝗵𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗧𝗲𝗺𝗽𝗼𝗿𝗮𝗹 𝗠𝗼𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝟯𝗗 𝗥𝗲𝗰𝗼𝗻𝘀𝘁𝗿𝘂𝗰𝘁𝗶𝗼𝗻
Songyan Zhang, Yongtao Ge, Jinyuan Tian ... Chunhua Shen
arxiv.org/abs/2504.05692
Trending on www.scholar-inbox.com
Songyan Zhang, Yongtao Ge, Jinyuan Tian ... Chunhua Shen
arxiv.org/abs/2504.05692
Trending on www.scholar-inbox.com
新たなデジタルマップ作成を楽しむ「iHistory FREE」プランが登場！#東京都#港区#デジタルマップ#PointMap+#iHistory
株式会社iHistoryが「PointMap+」に新しい無料プラン『iHistory FREE』をリリース。個性的なデジタルマップを作成でき、気軽にクリエイティビティを楽しめる新サービスです。
株式会社iHistoryが「PointMap+」に新しい無料プラン『iHistory FREE』をリリース。個性的なデジタルマップを作成でき、気軽にクリエイティビティを楽しめる新サービスです。
新たなデジタルマップ作成を楽しむ「iHistory FREE」プランが登場！
株式会社iHistoryが「PointMap+」に新しい無料プラン『iHistory FREE』をリリース。個性的なデジタルマップを作成でき、気軽にクリエイティビティを楽しめる新サービスです。
news.3rd-in.co.jp
April 14, 2025 at 11:36 PM
新たなデジタルマップ作成を楽しむ「iHistory FREE」プランが登場！#東京都#港区#デジタルマップ#PointMap+#iHistory
株式会社iHistoryが「PointMap+」に新しい無料プラン『iHistory FREE』をリリース。個性的なデジタルマップを作成でき、気軽にクリエイティビティを楽しめる新サービスです。
株式会社iHistoryが「PointMap+」に新しい無料プラン『iHistory FREE』をリリース。個性的なデジタルマップを作成でき、気軽にクリエイティビティを楽しめる新サービスです。
There’s room to improve. Future work could explore:
• Extending to in-the-wild scenes via foundation models for robot segmentation & keypoints.
• Ensembling predictions over time for better temporal consistency.
• Fine-tuning pointmap models on real robot data to handle cluttered tabletops.
8/n
• Extending to in-the-wild scenes via foundation models for robot segmentation & keypoints.
• Ensembling predictions over time for better temporal consistency.
• Fine-tuning pointmap models on real robot data to handle cluttered tabletops.
8/n
April 23, 2025 at 11:50 PM
There’s room to improve. Future work could explore:
• Extending to in-the-wild scenes via foundation models for robot segmentation & keypoints.
• Ensembling predictions over time for better temporal consistency.
• Fine-tuning pointmap models on real robot data to handle cluttered tabletops.
8/n
• Extending to in-the-wild scenes via foundation models for robot segmentation & keypoints.
• Ensembling predictions over time for better temporal consistency.
• Fine-tuning pointmap models on real robot data to handle cluttered tabletops.
8/n
Thanks Zhenjun for sharing. Our PM-Loss tackles low geometry quality in feed-forward 3DGS caused by "discontinuity" in predicted depth, and features a plug-and-play training loss that leverages geometry priors from pointmap-based models like VGGT. More at aim-uofa.github.io/PMLoss
June 9, 2025 at 6:26 AM
Thanks Zhenjun for sharing. Our PM-Loss tackles low geometry quality in feed-forward 3DGS caused by "discontinuity" in predicted depth, and features a plug-and-play training loss that leverages geometry priors from pointmap-based models like VGGT. More at aim-uofa.github.io/PMLoss
新感覚のデジタルマップ体験イベント、東京で開催！#iHistory#デジタルマップ#PointMap+
スタートアップiHistory Inc.が開催するデジタルマップイベント。新次元の体験を五感で楽しもう！
スタートアップiHistory Inc.が開催するデジタルマップイベント。新次元の体験を五感で楽しもう！
新感覚のデジタルマップ体験イベント、東京で開催！
スタートアップiHistory Inc.が開催するデジタルマップイベント。新次元の体験を五感で楽しもう！
tokyo.publishing.3rd-in.co.jp
March 24, 2026 at 9:30 PM
新感覚のデジタルマップ体験イベント、東京で開催！#iHistory#デジタルマップ#PointMap+
スタートアップiHistory Inc.が開催するデジタルマップイベント。新次元の体験を五感で楽しもう！
スタートアップiHistory Inc.が開催するデジタルマップイベント。新次元の体験を五感で楽しもう！
スタートアップiHistoryが新機能「myMapStore」を発表！デジタルマップ販売の新時代開幕#東京都#港区#PointMap+#iHistory#マッパーカルチャー
iHistory Inc.が新機能「myMapStore」をリリース。ユーザーは自作のデジタルマップを販売でき、クリエイターエコノミーの新たな可能性が広がります。
iHistory Inc.が新機能「myMapStore」をリリース。ユーザーは自作のデジタルマップを販売でき、クリエイターエコノミーの新たな可能性が広がります。
スタートアップiHistoryが新機能「myMapStore」を発表！デジタルマップ販売の新時代開幕
iHistory Inc.が新機能「myMapStore」をリリース。ユーザーは自作のデジタルマップを販売でき、クリエイターエコノミーの新たな可能性が広がります。
news.3rd-in.co.jp
June 2, 2025 at 3:19 AM
スタートアップiHistoryが新機能「myMapStore」を発表！デジタルマップ販売の新時代開幕#東京都#港区#PointMap+#iHistory#マッパーカルチャー
iHistory Inc.が新機能「myMapStore」をリリース。ユーザーは自作のデジタルマップを販売でき、クリエイターエコノミーの新たな可能性が広がります。
iHistory Inc.が新機能「myMapStore」をリリース。ユーザーは自作のデジタルマップを販売でき、クリエイターエコノミーの新たな可能性が広がります。
iHistory to Run Kyoto Life Demo on Apple Vision Pro at Tokyo Forum
Kyoto City's cooperation is what separates PointMap+Life from a tourism demo. A city letting a startup map daily life, not landmarks, is the part worth watching.
Kyoto City's cooperation is what separates PointMap+Life from a tourism demo. A city letting a startup map daily life, not landmarks, is the part worth watching.
September 10, 2026 at 2:34 AM
iHistory to Run Kyoto Life Demo on Apple Vision Pro at Tokyo Forum
Kyoto City's cooperation is what separates PointMap+Life from a tourism demo. A city letting a startup map daily life, not landmarks, is the part worth watching.
Kyoto City's cooperation is what separates PointMap+Life from a tourism demo. A city letting a startup map daily life, not landmarks, is the part worth watching.
新たな“暮らす京都”をデジタル体験！9月13日に特別イベント開催#PointMap+#iHistory_Inc.#京都市移住
空間コンピューティング体験会が9月13日に東京国際フォーラムで開かれ、参加者は「暮らす京都」を一人称視点で体験出来るチャンス！
空間コンピューティング体験会が9月13日に東京国際フォーラムで開かれ、参加者は「暮らす京都」を一人称視点で体験出来るチャンス！
新たな“暮らす京都”をデジタル体験！9月13日に特別イベント開催
空間コンピューティング体験会が9月13日に東京国際フォーラムで開かれ、参加者は「暮らす京都」を一人称視点で体験出来るチャンス！
kyoto.publishing.3rd-in.co.jp
September 10, 2026 at 12:07 AM
新たな“暮らす京都”をデジタル体験！9月13日に特別イベント開催#PointMap+#iHistory_Inc.#京都市移住
空間コンピューティング体験会が9月13日に東京国際フォーラムで開かれ、参加者は「暮らす京都」を一人称視点で体験出来るチャンス！
空間コンピューティング体験会が9月13日に東京国際フォーラムで開かれ、参加者は「暮らす京都」を一人称視点で体験出来るチャンス！
tl;dr: a subset of anchor images->VGGT->neural scene representation; this neural scene representation+all remaining images->VGGT with attention mask->pointmap+pose
August 26, 2025 at 10:37 AM
tl;dr: a subset of anchor images->VGGT->neural scene representation; this neural scene representation+all remaining images->VGGT with attention mask->pointmap+pose
tl;dr: intrinsic-invariant (/scale-invariant) pointmap+position-interpolated rotary positional encoding
July 23, 2025 at 10:16 AM
tl;dr: intrinsic-invariant (/scale-invariant) pointmap+position-interpolated rotary positional encoding