@JohanWinn Currently, it is lazy per timepoint (and somewhat per channel) which can reduce the size of at least some large timeseries datasets. I think any xyz laziness requires changes to the volume rendering engines, which is the plan from the blender dev team, but I won't do myself 😅
@JohanWinn But i also don't know for sure whether it's as desirable as in 2D views - when zooming in on the outside of a cell with a perspective lens, it is hard to know how deep to load the high-res data + you need to swap out a lot of data often
@JohanWinn Currently i think it's not easy, as I'm not loading directly from zarr, but resaving locally to the vdb format that blender uses (which can itself have pyramidal structure but is not currently lazy in Blender)
And with OME-Zarr we can easily load data from public databases on the internet, such as the Bioimage Archive or IDR. Here is a beautiful mitosis dataset from Walther et al. (2018), hosted at https://t.co/YJ9Y06tq9U
And with OME-Zarr loading, due to the pyramidal format, you can design your video at small-scale and only replace to full-scale at the final render!
(data from Granita Lokaj, @ https://t.co/GmpGsGnuY8)
@StanwadeWang And it's good to know that the .vdb FloatGrid that encodes the volumes is 32-bit depth values, so this may add some extra pressure on your memory 😅😅
Version 1.0 of tif2blender is online! This includes lots of new features 🥳🔬
To use this, see the brand-new youtube tutorials! These take you from installing, to loading to presentation-ready video for fluorescence and electron microscopy!
https://t.co/zxptbl4gzY
@StanwadeWang The limiting factor is currently (i believe) just RAM and render time and responsiveness. If your 50gb tif has many segementation channels or multiple channels you can turn off it's very doable, otherwise i would consider downsampling (as you also usually render to max 4k pixels)
For the EM data, this is a dinoflagellate FIB-SEM stack provided by @MocaerK. Here we show the EM stack as a volume render, with animated slicing that is separate for the segmentations and the data. EM tutorials: https://t.co/67uHoZgCMd
For the fluorescence imaging, I'm using C. elegans meiosis data gathered by Ivana Čavka, annotated with BigTrace and SpotMax.
Here we show the 5D (xyz+channels+time) with annotations in Blender!
Fluorescence tutorials: https://t.co/fTVNeO02ON
🔬 1/6 New paper and software – hot off the press! 🐾 Introducing OmniTrax, a new open-source @Blender Add-on designed to streamline multi-animal tracking and pose-estimation. Check out our JOSS paper: https://t.co/wLujQq7fAh
The benefit/downside of live imaging is that sometimes one observes too rare or strange events (to be included in a publication).
But Twitter is ok: a LifeAct labeled COS-7 cell that decided to eat something nasty from a medium, regretted it, suffered, but found a way to recover
Great piece in @Nature today exploring if its too late to keep warming below 1.5C (yes, at least without overshoot) but also highlighting the accelerating energy transition that is occurring worldwide. I was happy to help with a few figures in the piece: https://t.co/c1pOJLU5tm
@katpyxa While Cycles (the #b3d raytracing engine) does do raytracing through the volume, but with cutoff values for the number of bounces/depths that rays can traverse and some other optimizations (refracted rays do not continue into the volume, for example)
@katpyxa Very interesting thread! As far as i understand (but I still haven't committed to going into the #b3d and openvdb source code) blender in eevee/openGL samples volumes at different depths from the camera, to then be able to participate in openGL z-sorting for transparency
Blender Foundation and the online developers community are proud to present Blender 4.0 🚀
Light Linking, AgX, new Principled BSDF, Geometry Nodes Repeat Zone, USD improvements, and so much more!
See all that's new at https://t.co/PcXpBKul9A #b3d#DevFund