All of my 3D this year was created by AI -- not just the printable objects, but also the AR (Augmented Reality) and Hologram images.
3D SOFTWARE
The AI 3D Programs that I used this year:
- POINT-E (no longer working)
- SHAP-E (no longer working)
- LATENT LABS (no longer working)
- CSM (Common Sense Machines) (on Discord)
- GENIE from Luma Labs (on Discord)
- MASTERPIECEX
- MOOTION
- LEONARDO.AI (for texturing)
FAKE SCULPTURE EXHIBITIONS
In the spirit of fake 3D, I used AI with Midjourney to remix the 2023 exhibition of O'Keeffe and Moore at the Albuquerque Museum, to make a new fake exhibition of sculptures:
I'd still love to have a fake exhibition -- by placing my digital 3D creations in the Sculpture Garden at the Albuquerque Museum with AR (Augmented Reality):
My AR fantasy turned out differently in Santa Fe, than I had imagined below:
3D P R I N T S
2D Image to 3D Print
The ULTIMATE satisfaction in 2023 came when we twice 3D printed -- an object that was generated in CSM, based on one of my 2D drawings:
- First, Tim printed a one inch figure from the 3D resin printer at Quelab
- Then Adam printed a 7.5 inch PLA figure remotely, from Quelab on the Bambu 3D printer at his house (which he delivered to me on December 12th)
Both the large and small 3D prints can be seen in the photo below:
DOWNLOAD: "ropesculpture.GLB" (3.81 MB)
3D PROGRAMS
The 3D programs I used in 2023 are listed below, in chronological order of how I discovered them:
POINT-E
BOCCIONI 3D
At Quelab we 3D printed a Boccioni sculpture that I had distorted (how post-modern of me), and then regenerated in 3D using POINT -E (which debuted in late December 2022):
In 2020 I had downloaded and distorted the famous sculpture by Umberto Boccioni -- Unique Forms of Continuity in Space-- using SculptGL.
I then broke the 3D file that I had created, down into selected 2D views, to make the animated GIF below:
This year 2023 I uploaded the simplest Boccioni 2D view to POINT-E, and generated another 3D object file -- the one that Tim 3D printed at Quelab. The image below shows the original sculpture first, followed by the distortion I made in 2D, and ends with the tiny 3D print we made of my distortion:
DOWNLOAD Boccioni_cleaned-up.STL (224 KB)
Previously Ander Raden and Matt Smith in the Netherlands used technology (but not AI) to recreate four lost Boccioni sculptures and exhibited them in 2019:
Last year, in 2022, I further distorted this Umberto Boccioni sculpture with AI using Midjourney:
Midjourney prompt: "in the art style of futurism"
3D JAGUAR
Image-to-3D
Below is the interface of Point-E text-to-3D Demo, for converting a 2D image into a 3D object:
DOWNLOAD Jaguar_fixed.STL (622 KB)
VIDEO
of 3D Leonardo.AI Sculptures
After generating a bunch of sculptures from my drawings, with AI using Leonardo.AI, I converted them into 3D objects with POINT-E. Then I imported the 3D files into Blender and made a video:
- BLENDER file: 230225__AI_sculptures.blend (2.67 MB)
- Sculpture 22: 22__sculpture.obj (215 KB)
- Sculpture 36: 36__sculpture.obj (213 KB)
- Sculpture 37: 37__sculpture.obj (853 KB)
- Sculpture 50: 50__sculpture.obj (213 KB)
SHAP-E
Text-to-3D and Image-to-3D
On May 12, 2023 Open AI came out with an improvement over POINT-E -- another 2D-to-3D program called SHAP-E.
So I uploaded a 2D image that I generated in Midjourney, with a lot of shading, to SHAP-E, and it did a better job of generating a 3D object:
I then added "texture" and color -- "green algae covering the shapes" -- to the 3D SHAP-E result using the Leonardo.AI "Texture Generation" feature:
DOWNLOAD: green algae covering the shapes.zip (32.80 MB)
LATENT LABS
Text-to-3D
On May 23rd LATENT LABS released a Text-to-3D program in Discord:
I asked for a "javelina," and the result was just ok:
DOWNLOAD: javelina.OBJ file (2.63 MB)
CSM
2D Image-to-3D
On August 3rd I got access to CSM (Common Sense Machines), a 2D-to-3D program on Discord. I think this is the best AI 3D generator of 2023:
I then uploaded a female figure generated by Midjourney, based on one of my drawings, and CSM generated four preliminary views in Discord:
Then it took a while to generate a full 3D file in GLB format:
DOWNLOAD: the GLB file (3.65 MB)
I uploaded a second Midjourney figure based on my drawing, generated a GLB file in CSM, and 3D printed it at Quelab:
The 3D print of the above second figure:
DOWNLOAD: the GLB file (3.98 MB)
CSM works decently/well with a more developed 2D inputs:
DOWNLOAD the 3D files of the above figure
DOWNLOAD: Notwhistlersmotherchair.GLB (3.7 MB)
DOWNLOAD: "KrrrlBernini.GLB" (3.35 MB)
DOWNLOAD: "Portrait_CSM.GLB" (4.24 MB)
I wasn't happy enough with the result
to offer it for upload,
because it was lacking texture
Download: "231128__leonardoai.GLB" (3.67 MB)
POLYCAM
Gaussian Splatting
POLYCAM Gaussian Splatting has a free trial right now, where one can scan his environment and their software will generate a 3D file. I uploaded a turntable figure and got the mess below, as seen in Blender:
I believe that one needs a background in all the images in order to successfully generate a 3D file, probably because the program uses parallax. Images of a turntable figure (like the one below) probably don't work.
GENIE
Text-to-3D
On November 1st, LUMA LABS released GENIE, a text-to-3D program on Discord:
I generated a decent "robot javelina":
DOWNLOAD: Javelina_Robot.GLB (4.95 MB)
Later on I generated a couple of decent "alien javelinas":
DOWNLOAD: alien javelina__highrez__04.GLB (3.03 MB)
- DOWNLOAD: alien javelina__highrez__03.GLB (3.10 MB)
MASTERPIECEX
In November discovered MASTERPIECEX, which generates an animated 3D file from text (but not from 2D images):
DOWNLOAD 3D Animation:
An_artist_with_a_bad_haircut_1.GLB (18.83 MB)
MOOTION
In December Matt Wolfe's YouTube told me about MOOTION, which generates animated 3D files in FBX format, as well as videos:
DOWNLOAD: back_flip_2_bot.FBX (2.79 MB)
WHAT NEXT?
Maybe Quelab needs a 3D printer that will print in clay, like the one they have at the Hand and Machine Lab at UNM:
H O L O G R A M S
The Looking Glass Portrait Display is another way to show art in 3D, especially considering that the 3D files can be converted into Holograms:
The Looking Glass Portrait Display in action, showcasing a Deforum video from different angles:
DOWNLOAD: Krrrl_drawing_clothed_quilt.jpg (4.63 MB)
HOLOGRAM Software
- Alice/LG 2.2 -- Blender Add-on (YouTube tutorial) -- for converting 3D files into holograms
- The simplest way to make a Looking Glass Hologram is to use a depth map, but my previous programs no longer work
HOLOGRAMS to DOWNLOAD
I converted a lot of my YouTube slideshows of AI images into Hologram slideshows, as noted below:
53 "HOP" files can be downloaded:
Hologram QUILTS
from videos
%20(1).gif)
DOWNLOAD: Barbenheimer quilt (3.66 MB)
DOWNLOAD: Galloping Horse quilt (4.33 MB)
These 3 online programs no longer work -- but they would give me the DEPTH MAPS that the Looking Glass Portrait display needs to generate a 3D HOLOGRAM:
A U G M E N T E D REALITY
Besides holograms, one can display 3D files in AR (Augmented Reality):
Software
- WebAR (upload and test AR)
- WASM AR (recognizes a "target" image, like a QR code, and then conjures the AR automatically) (YouTube tutorial)
AR Links:
I prefer using WebAR, since people do not need to download an app to view the Augmented Reality. Adric programmed an NFC card with the AR Tapir link, and we put that behind one of the AI prints at Quelab. Then one would only have to wave his smart phone close to the lower right hand corner of the print to conjure up the flying Tapir in 3D:
We also experimented in AR with one of my 3D drawings made in Paint 3D:
I enhanced my print for the Southwest Print Fiesta Print Exchange 2023 with the AR Tapir. Grant used WASM AR to scan my print and conjure up the 3D AR tapir in WebAR, with a smart phone. However the process was really painful, with too many steps, and not-ready-for-public-consumption:
First one would have to scan the QR code below, to get to the WASM AR site: tinyurl.com/KRRRLTapir:
PREVIOUS AR
We have previously experimented with AR successfully in the past:
In 2015 three of the Desert Triangle Print Carpeta prints were enhanced with AR:
We 3D scanned the big Xerocraft skull, and 3D printed it at Xerocraft hackerspace in Tucson. We placed that skull 3D print on an AR "target" and viewed the AR skull through a tablet, during the Tucson Sculpture Festival 2013:
Spinning AR! The AR "target" and 3D printed skull were on a lazy Susan, which could be controlled with a smart phone, which spun around the the AR and physical skulls:
I still want to create sculpture that can be controlled with a smart phone!
******
What if this whole AR Print fantasy could be streamlined in real time? One would hold their smart phone over the framed 8x8 inch print, and the phone would take a picture of it automatically, like it were reading a QR code. That image would then be sent to the cloud, where Shap-E would automatically generate a 3D image. Then that generated 3D AI image would then be projected over that same print in Augmented Reality, practically in real time, adding a 3D element to a 2D print.
Moreover the smart phone could deliver an older 3D AR image immediately (perhaps generated by the previous viewer), while the new 3D AI AR image was being generated.
One might even use Siri, or another voice command assistant, to influence the 3D image that Shap-E generated.
ARHero.io is going in that direction:
No comments:
Post a Comment
Note: only a member of this blog may post a comment.