There are definitely some things I just read in there that are inspring to me and I think I am on track to do some of the things mentioned in that article. Thanks for sharing that, even though it’s off topic I think, still an insightful read!
I do remember an early forum member Tomsdesk creating excellent trees, 2D 2.5D. Could this be the ‘someone’ you refered to?
“https://groups.google.com/u/1/g/sketchupgallery/c/yE5joDLSrQU/m/bKMyXL0uEJAJ”
Hey Dan, the link you shared didn’t work to access on my end!
No. I do remember Tomsdesk’s trees, they are excellent. My remembrances might even be pre-Google, and they were about a very NPR way of presenting.
I have those. I agree. they are nice in both SU and NPR renders.
I wonder about abundance though @Anssi. They are nice for sure, I like the most realistic looking symbols, also there seems to be a lot more images of trees than pastel looking 2d trees like with what tomsdesk has… Anyone have good tips on making the 2d realistic tree photos look more subtle? I’m guessing adding a bit of transparency and maybe turning the saturation down a bit could be a good start? @AK_SAM , what would you do to make a realistic tree photo 2d object look more subtle, tranparency and maybe some saturation changes?
One thing I’d also like to do, is to not have a TON of symbols for top view plans, I’d like to just pool similar trees to be on one tree symbol, same with shrubs, have a few different colours and sizes for variety but not have a specific symbol for every unique kind of vegetaion, which I don’t think many people do anyways, there are too many types of vegetation…
While I do have workflows for 2d cutouts/etc, I actually rarely use 2d images within a scene.
I use 2d images sometimes in elevations or cross sections, in a quite a symbolic manner (sillouette style; same as people, i dont use cutout people).
I some times use 2d cutout objects them on very large scale Urban Planning projects, and then use the artistic effect like gsharp does.
I never use 2d “cutouts” in proper 3d renders. if you have 3d trees, why use 2d?
People that use 2d PNG stuff tend to be low-level arch-viz or landscape archtitects and they tend to compile everything in Photoshop with some artistic effects to blend it all. Personally i find that a very laborious and rigid way of working…it’s not 3d modelling, it’s more artistic illustration.
For 2d Plans, I use line symbols (2d+3d components with tags, same as windows). Group them as you suggest, eg Evergreen as one symbol type, Decidious a different type… you dont have to be too rigid about that. You can add colour to symbols to give variety in the plan view and add annotations (ideally tags, same as window schedules).
Once you’ve compied figured out all this advice you’re receiving, why dont you make a new post containing all your fresh ideas and optimised workflow? It would probably help a lot of others.
The question has come up in discussion- What are the Pros an Cons of 3d vs 2d (“faceme”) components?
Here are my thoughts/experiences:
First , it’s important to note that everybody will find their own style and workflow…part of rendering is experimenting and learning what *you like the look of.
To clarify: 2d ('faceme") components are usually photos of trees (or people, or whatever) which are “photorealistic” of course. So they look great ! I’ll use trees as the demo since that’s the most common type, but this applies to just about anything.
But they have some pretty clear limitations:
-
As soon as you tilt the camera to a birds eye or Plan view, the 2d-ness of the tree becomes apparent.
-
As soon as you change the shadow angle, the 2d component may look out of place because light/dark areas won’t look natural or consistent with your scene. Many 2d trees don’t have a strong shadow angle so they don’t convey much “depth” of lighting.
-
As soon as you add a second tree of the same type, you will have two identical 2d “cutouts”. You can flip and scale them for variation, and apply some colour adjustment, but that only works for very similar looking objects (like Pine trees). Good artists use multiple images of the same species of tree or plant, to give variation. (with 3d trees, its much easier to rotate and scale to add some very convincing variation to each omponent, while sketchup sees them as being identical components)
-
When you animate or change your camera view, (eg sharing in Trimble Connect or rendering in Twinmotion/Enscape) a 2d component looks the same in multiple rendered images (one from the front of the house, one from the back…but the trees are showing the “front” in both images).
-
When one tree is placed behind another or placed in among other trees/objects a 3d tree will have bits of light and shadow falling across its branches and leaves it in a natural way.
-
Renderers bounce light from objects and that reflecting light “spills” onto nearby objects - so a so a white wall next to a green tree should take on a greenish hue - that wont happen if using 2d components.
If you are prepared to accept all these limitations, then 2d components are fine and can look realistic and are quick to import and keep SketchUp file sizes small.
I find that they are best used sparingly or for specific types of details eg a plane flying in the sky, or a flag in the background. I also sometimes use 2d Silhouette people , because I like the effect.
Good artists using 2d images (whether in photoshop or 3d modeling tools) will generally build up a library of plants, trees, people and cars, with each example (species of plant) taken from different angles and even lighting conditions. I used to search my city for trees take several photos of them at different angles, then cut them all out in Photoshop (using alpha channels and transparency masks…it needs to be done well).
On the other hand, 3d components can take hours to model one specific tree, or otherwise be expensive to purchase and difficult to find each species you need. If you do learn to model trees then it gives you the ability to model specific trees in projects EG, a 200 year old Oak Tree that the client wants to retain.
Basic (low detailed/low face count) 3d trees can be used if your lighting and scene “depth” is important (in architecture/landscape design, sometimes conveying the true sense of SPACE is actually more important than the look of individual objects).
Detailed (photorealistic) 3d trees take 20,000 faces or more each, so you need a strategy for working with large file sizes and sending only the critical information to LayOut (nobody I know actually dimensions to a branch or leaf…so you shouldn’t be continually exporting an updating those details in LayOut.
Photo of a tree on a fairly overcast day: Could be convincing in a scene if this were a 2d cutout. You would have to make your time of day and shadow angle as neutral as possible…eg overcast.
Compared to a photo of a tree on a sunny day - this looks a lot more “dynamic” due to the lighting.
Then of course you have other uses for 3d objects, including
Night time scenes/lighting:
And Scattering:
Polygon counts maybe an issue of the past with EPIC’s new graphics rendering technologies
Nice tech demo here from 2010/2011.
Essentially - your screen has X number of pixels so in theory you should be able to have any amount of data behind the scenes , eg a trillion pixels, and your PC only has to deal with the 4,000 or so that you need to see at one time.
A bit like with Google Earth how it contains every detail of the whole planet, but you’re only streaming a “screen’s” worth of detail at one time which makes it very fast.
This was incredible to watch.
Well you bring up some really critical points in my mind. Your answer mostly answers the 2d vegetation thing for me… I’ve been trying to solve this for days, whether or not I wanted to use 2d trees in my workflow and how much I want to use them… You bring up a lot of good points I didn’t notice. Thanks Sam. This is turning out to be quite the comprehensive thread on landscape best practices using Sketchup…
For anyone wondering about the number of edges and faces a high quality tree should have for great rendering, it can be for example around 2000 faces and 20-30k edges. @AK_SAM had mentioned in a post above that a good quality tree for great rendering starts at 20,000 faces, that should be roughly what the edges are. My two cents anyways…
Depends on the tree of course, but that’s a ballpark what I just mentioned. At the same time, I have seen a good low poly tree with similar amount of faces that Sam mentioned… Anyways just thought I’d chip in, I think that number for the most part should be a lot lower than 20,000 faces. Say around 5 thousand… I say this is the number range roughly after just now doing some enscape closeup rendering tests, the tree looks amazing in enscape with that amount of poly. Lookup Dynascape tree on the 3d warehouse for an example and send it to your renderer, you can see for yourself.
If you want more accurate ideas, sketch3d.dynascape is probably the best for low poly veg that looks great, it says the poly right there on each item, so check those out for references. Rendering tests should be done too before buying a bunch of stuff so you know what you’re needing and buying is worth it.
WOW!! This is evolving at a fast pace. One wonders what is next.
Yep, wow! Real time rendering may be breaking out from that typical “Game Engine” look that I never really liked for architectural design work.