Open josuemtzmo opened 3 years ago
FYI, this will generate in total 3 different videos, one for each field.
In my opinion, since these are to serve are examples, then they are absolutely fine as is! What should be clear, though, is a comment somewhere in the scripts (or wherever appropriate) informing the users what they should edit to change the camera focus to their liking.
(@josuemtzmo, perhaps you need to ping some people if you want their feedback?)
@aekiss @AndyHoggANU @adele157; Do you have any comments on how to improve these stills before I commit to render the animatios?
I agree with @navidcy that they are fine as examples, and the main thing is to include comments to help people learn. But I also like your suggested changes @josuemtzmo. Perhaps they could be used as part of a how-to guide?
Also, some small suggestions:
You can change the camera view on one of them...
OK, but these are not just for examples, right? We also want to create some animations that we will use as part of this project.
@AndyHoggANU is this directed to @josuemtzmo ? Because I don't understand exactly what you mean. But I don't need to (if it wasn't been addressed to all)! 😀
Salinity animation: https://youtu.be/A6S2w-Hleho
Is this just an announcement? Invitation for review?
Invitation for review, and comments if you have any :)
There is no contrast... I almost can't see anything. But hopefully this will come with some documentation so people can modify the colorscale??
I have to say I don't have any background of what's happening here. Is this gonna be just a collection of animations to showcase the model? A repo that helps people make animations? Bit of both?
So, actually, the original aim was for Josué to produce animations that show off the model. But, of course, we want things to be reproducible, hence the repo where people can borrow code to replicate if wanted.
But we should try to perfect these simulations. I agree with Navid - the colour scheme is quite washed out - this is fine in the Eastern Indian Ocean where there is good contrast from the lighting, but everywhere else the lighting is too flat. Is there a way to get shadows over the whole domain? If not, we might need a stronger colour scheme?
Incidentally, the stills from the other day had much more contrast. Maybe consider eliminating the second light source again?
Incidentally, the stills from the other day had much more contrast. Maybe consider eliminating the second light source again?
Yeap, the stills looked nice.
Ok, I will first make the second light dimmer, and if we don't like it still, I will completely remove it.
Ok, I will first make the second light dimmer, and if we don't like it still, I will completely remove it.
Why not make it exactly as the still previews above?
I thought the rapid shift to the black background was too abrupt, but it sounds like you really liked the previous version, so I will remove the light then.
What we want (I feel) is to be able to distinguish flow features! It's not the contrast with the background but rather the contrast between regions with different salinities.
In general @josuemtzmo I feel you like a lot these pale colormaps which make it harder to visualise things. I'm wondering if to your eyes they look different and not as "washed out". This last one may be more of a philosophical question since it touches on perception of vision by different people... something that goes beyond the scope of the repo I feel.
If I had the chance I'd reduce the color limits to, e.g., 32-35 in an attempt to enhance contrast. But that's just a thought -- I don't know exactly how that will look like.
Here are some options:
When you give me too many options I can't select anything.
I'm sorry to admit, but for my eyes the same issue holds: colours are too washed out. I may be a singularity...
Still I think the salinity version here is better. The color limits are reduced there. Why not reduce even more?
I think, either option 1 or 3, looks perhaps even better than option 4 (Previous version). Fair point, I missed that I had selected a range of 30-36.
make it even smaller!
OK, I kind of agree with Navid's comments ... but I also think that the nice think about your images is the relief, which highlights the gradients. But on a globe you will always have regions with no relief. This is easy to address when you are zooming into a region but is there any way to fix it for a global picture?
I can increase the relief, but at some point, it will look overexaggerated.
I've reduced the color range, what do you think?
In the image below, I've exaggerated the relief, but It starts looking strange to me.
The image at the bottom barely starts looking nice to me!
Correction: The one at the bottom actually looks nice!! I'd like a better colormap to help my vision but I don't think I would be able to get there unless I make the animations myself. I'm hoping that at least the directions that will be added will help me choose the colormap of my liking. Until then, I'll keep quite since there is no point repeating myself. 😌
With some annotations:
I think the color range is still too wide... Or I don't know, there is something that obstructs my eyes to see any contrast... the choice of colours... I'm not sure what it is! The problem is that @josuemtzmo doesn't see this. (I'm intrigued by this difference in visual perception btw...)
It's not a "problem" really... it's just interesting!
I was using the cmocean haline
colormap, and there is no much I can change of it, I've tried with a different colormap viridis
, which may highlight a bit better that side of the sphere. I need to point out that it's a 3d object and the light reflection makes some of the colors wash out. I've removed the sphere reflection, and perhaps that will improve the contrast...
I need to point out that it's a 3d object and the light reflection makes some of the colors wash out. I've removed the sphere reflection, and perhaps that will improve the contrast...
I understand, but I believe we want to be able to visualise flow features. We aren't interested on 3D visualisation just for the sake of 3D visualization, right?
@josuemtzmo just render a movie and it'll be fine. Don't listen to me... Document the script that you've used to do so and when I would like to make an animation for some work of mine I'll tweak it to my liking.
[I will stop at this point because I am already afraid that I've become obnoxious.]
@navidcy It's fine, I will document this, but the idea is to have something that most people will use, although it will be possible to reproduce the animation, It's not as trivial as just running a script.
I understand that the contrast is not so high, where the light directly hits the sphere, I will keep playing with it, but there are so many options for the lighting that I don't know which combination will fit the best, for example, the sphere is lighted by the color of the image projected in its surface.
@navidcy Do any of the images above fit better your preference? So I can narrow down my exploration... @AndyHoggANU, do you have a preference? or should I keep exploring?
This last one is looking better. Still a bit washed out on the right, but good relief elsewhere. How did you do that one?
I think it is good to get these comments, Navid. Josué can do all the technical bits, but it is hard for him to judge the output.
I'm a bit reluctant to whether I am crossing the boundary between making remarks on objective visualisation attributes and on subjective personal taste and preferences.
BTW, did you ever produce the video we discussed with the ice + temperature, or ice + vorticity? I think the plan here was to do a lap of the Antarctic continent.
I prefer viridis
to haline
- it is good to have more hues to make salinity gradients more obvious.
If we are happy to focus on spectacle rather than perceptual uniformity/accessibility we could increase the number of hues still further by committing a blasphemy like some sort of rainbow, eg turbo
, nipy_spectral
, gist_ncar
. These would make salinity gradients more visible and therefore make the flow features stand out better, at the cost of a less accurate perception of the salinity magnitude. But perhaps the former is more important than the latter for this application?
Also the mesoscale flow features are quite small on screen and easily overlooked - could we zoom in as part of the animation?
What would it look like to have different coloured light sources from different angles? Or would that be too confusing?
@josuemtzmo a suggestion: why don't you bring this to a group meeting and we do the modifications + comments live all together? Then you don't have to go back and forth the github comments and the drawing board. That'll be fun, no? Or isn't quick enough to be doing such modifications online and showing to all live?
by committing a blasphemy like some sort of rainbow, eg...
Blasphemy... 🤣
Also the mesoscale flow features are quite small on screen and easily overlooked - could we zoom in as part of the animation?
@aekiss I can definitely zoom anywhere, do you have somewhere in mind?
I need to point out that it's a 3d object and the light reflection makes some of the colors wash out. I've removed the sphere reflection, and perhaps that will improve the contrast...
This animation looks 3D but yet I wouldn't say that colors washed out. I actually find it has enough contrast to be able to see the flow features, be quantitative, and pleasing to my eye.
I guess just in general if it was magnified by (say) 2-4x it would show up the mesoscale much better. Some interesting places are the ACC, ASC, EAC, Leeuwin.
This is a sample animation with the latest viridis
colormap:
We could definitely have a meeting to modify things live, but there are many options and the renders take up to 1 minute per frame.
We could definitely have a meeting to modify things live, but there are many options and the renders take up to 1 minute per frame.
I was thinking just for just still images over group meeting. But 1 min for every change may be cumbersome...
Ok, perhaps I should try some outrageous colormaps... (I was trying to avoid it) For example, the one @navidcy sent.
@navidcy - nice example, with a blasphemous colourmap!
Unfortunately @josuemtzmo, it's more important to be able to see something than have a colormap that is proper and washes out all flow features. I'm not sure if it's a colormap issue, you are the 3D visualization expert. We are describing here what we would like to see, we are not describing solutions. We hope you know how to make our eyes satisfied.
✊🏼🤙🏼❤️
The following snapshots show MLD, salinity, and temperature from ACCESS-OM2 0.1 degrees.
MLD
Salinity
Temperature
The changes I have in mind are:
Please let me know if you would like any other change, otherwise, I will upload a sample video in the next couple of days.