Completely new to Blender, and doing my Master Thesis using Blender

[Mats_Ausland (Mats BA)]

Hi everyone! This is my first post ever, so please be patient with me!

As the title explains, I am completely new to Blender. I have finished BlenderGuru’s tutorial for a Donut, and that is it! However, I did use a program Called Cheetah3D before where I was trying to accomplish the same thing, the program got substituted with Blender by my teacher, and for about 2 weeks, I have been trying out different methods and try to learn the program as best as I can.

My story is that I am currently working on my master thesis which evolves around creating an AR app in the area around the City hall of Oslo. My task would be to modell the area around the City hall and the City Hall itself, in different timelines. I would like to showcase how the area looked in 1840-1930 before the construction of the City Hall, and then I want to modell the area when the construction began 1930-1950, and then how it looks now. Or an in-depth segementation of the construction era process.

As it stands I just finished the “How to create a city in Blender” on YouTube which showed me the insane level of complexity and opportunities that exists when working with Blender in contrast to Cheetah 3D. While the OpenStreetMap addon worked great, it didnt give me what i needed. The City Hall looked amazing, but the mapping for terrain and the satelite image overlay was from how it looks today.

So my question is, how should i best move forward when modelling this area, I will try to add some early references of the area through old maps, but i need a way to generate the models, or create them manually. Answers on both ends are appriciated.

In the Cheetah 3D forums i created a thread similar to this, where other users found me some blueprints of the City Hall itself. I used these blueprints in Blender aswell and tried to create the City Hall i Blender, and then i kinda got stuck. Appriciate any help anyone would give! And as I am brand new to the software i apologise in advance for the headache people may gets when responding to your threads! I will also attach the Cheetah3D forum link in this thread, so that additional information about my subject is easily accessible. Thanks!

New to Cheetah3D - Using it for master thesis. Help!

[Mats_Ausland (Mats BA)]

It would not let me add multiple pictures, so I will try here instead.

Top (As you can see, the one side is stretched forward, and i dont understand if that is really how it looks, or just a mistake.)

[Mats_Ausland (Mats BA)]

Front

[Mats_Ausland (Mats BA)]

Back

[Mats_Ausland (Mats BA)]

Left (Both Right and Left side are somewhat identical, just different details)

[Okidoki]

Do you even know how much you wann do? How big will this “area” be. I mean jsut have look at Carcassonne Medivial City for example.

[Mats_Ausland (Mats BA)]

Hey! Thanks for the reply! Yeah, the area should be as small as possible tbh, as few polygons as possible, seeing as this will be rendered though a smartphone.

The most important elements i would like to include is the “Vestbanestationen” and “Søgaden” which is the lines of houses across the center.

Here is a cropped version of the photo above.

[Mats_Ausland (Mats BA)]

Here is how it is today

[Mats_Ausland (Mats BA)]

Here are the houses right above “Søgaden”

So it should be quite easy to make for most of you guys I would presume! But what I seek is only some information on how you would attack this task! Should i begin with mapping out the spacing between each house, or just start modelling the houses and the city hall, and then scale them to size?

[pitibonom]

hi @Mats_Ausland

This looks like a nice project !
congratz for you might for doing this with a tool you don’t know perfectly : blender.

It’s pretty difficult to answer you in an accurate way and show you a path more than another, because IMHO the 1st thing to know is what do you want this final AR software to do ?
Will it have to run on a low-end 3D device like a smartphone with a decent frame-rate ?
Will the 3D place be very wide ?
Do the texture have to look realistic if the visitor gets close to the buildings ?
Will you have to show the VR in night and day ? or only in day ?

Answering this will tell you wether your models can have hundreds of thousants of triangles or not, wether you can embed 500MB or more of textures, and this is the only blender side things.

An important question you have to answer, if you want some realtime rendering is your choice of a 3D rendering engine.
Though most use to say they all offer about the same features they are all different in achieving things.
some use java, others python, others, c#, others use lua, etc… My choice went to unity3D for its 2 main features: versatility and documentation.

Finally, for answering your question, modeling the building shown in the blueprint should be quite straightforward as it’s mainly made of parallelepipedic parts. The gif that can be seen in you cheetah forum will be quite the same in blender:

• make a cube
• use loop cuts
• extrude some parts
• use modifiers to repeat cubes and arches for windows and openings
• use booleans with them…

this is just an idea that might not be the best one.

The idea to have underneath your blender models the map of the whole place is very good.
Though you don’t really care with individual models scales, the map will give you the overall models sizes, preventing you from going too far from reality.

The modeling part will be IMHO the fastest and easiest thing
Texturing ( though it can look quite straightforward ) and unwrapping is not this simple.
And if you want to replace textures with pbr materials, it will be even more difficult as you’ll have to play with light, nodes and textures…
So this all depends on your knowledge on 3D creation and if you are good at it ( not like me ) this might be quick and pleasurable to make.

At last but not least and just because the finality of your work seems to be an app, i’d suggest you to set-up the whole toolchain you’ll use, pretty early as from experience, i made amazing looking things in blender that looked ugly and shitty in unity.
So do a try and show it up in your final engine as quickly as possible. This will avoid you to rework lots of things 3D before your end milestone

please keep us informed of your progress here, i’m personally curious with what you’ll do

Happy blending !

EDIT: AH ! last but not least: NEVER use one and only blueprint for a big project. Blueprints are ( very ) often inaccurate. ALWAYS stack multiple images from various origins ( google maps, geoportail, cadastre, etc… ). You’ll see few wil match one another and only your eye and brain can do the proper compromise when modeling over those maps

[Mats_Ausland (Mats BA)]

Wow! Thank you so much for the answer! Well, it so happens that the engine I will be running it on is Unity. And I actually have a second programmer who will do all of the work in Unity, so I am not able to switch between Blender and Unity, as I will have to send it to him before he works on it. So I hope I dont make it difficult for him!

As for the AR part, AR is just a part of what the task entails. It will not be Augmented Reality as you may know it, but what my supervisor calls “Situated Simulations”, so I will only model within my restricted area, and what is modelled, does not really correlate with the “Reality” per se. It is locked in this simulation through the app. Here is a link to my supervisor home page. http://www.sitsim.no/ hopefully that answers some of the questions you had about the AR software. It will indeed run on a low-end 3d device, and the place looks to be kinda wide as well!

To be short, I do the modelling and the refrence points (according to scale, maps and info about terrain) and another co-worker of my supervisor will add it to unity and run it. My task is only to model each building using Blender, and to create these hyperlinks in Unity, which goes under the category of “Gamification”, where the thesis would tackle concepts like City Tours, or City Guides, and in a sense, compete with these concepts. Is it possible to extract knowledgable information through smartphones, and through the interactiv process of Augmented Reality?

Again, thanks for your answers! I really like the way of starting with a cube or a plane over the Blueprints, but I found it harder in Blender then Cheetah, just because i didnt have proper training in Blender. But I’m sure I will figure it out! Again, thanks!

[pitibonom]

Okay
i seek better what you need to achieve by looking at this vid:

Old Narva – A Photo Positioning Puzzle (Sitsim DEMO VI)

You seem to run all this on apple devices and thought i don’t like apple development ( even if i work on it for my Carcassonne medieval city project ) This will allow you lots of things as iphones and ipads offer insane 3D power and features.

I gues you should start with the base cube in edit mode, and play with the CTRL+R feature.
You’ll get lots of help from youtube tuts and vidz for various questions.
As you probably noticed, blender offers lots of keyboard shortcuts. You should learn them by practicing.
I guess there are as many blender keyboard setups as users
Mine is a mix between old features like the W key that exists no more in 2.92 and some features now drown in big menus that i added in my fast shortcuts.

The num keypad is important and the 5-KEYPAD is usefull for a parallel projection needed when you model over a blueprint.

some common shortcuts like G for Grab or GG for Grab-along-edge or s for Scale or R for rotate, associated with the X ( for global ) or XX ( for local ) key will grab, scale or rot only on X axis…

IMHO you should take a keyboard practice time. At 1st view you feel it have been written by morons for prehistoric people but you will quickly be unable to get rid of those shortcuts

Also you should get familiar with the ‘centers’. the 1 in my pic.

The most usefull ones for me are cursor ( the 3D cursor that you can place wherever you want ), individual origins and median point.
They are the center of whatever you do in blender in object or edit mode.

2nd are the cartesian domain you work with. global, local, normal. You can also create yours from a non aligned face that you need to use as base reference.

Mix of for example individual centers in 1 and normal in 2 allows you to extrude multiple faces independently from one another, towards their own normal. This is insanely powerfull !

The 3 is the snapping. I use always the vertex snapping and rarely the edge snapping. This is the mandatory feature that allows you to have 2 faces sharing an edge without a void between those 2 faces. The CTRL-KEY is your friend for doing this.
here’s my setup: i snap for move,rot and scale.

The 4 is the gizmos params. Some helpers shown in 3D to help you on what you do and where you are

For helping you, there are also overlays: 5 ( more interresting in edit mode )

Usefull if you need some edge/angle/face measurement, or see a face barycenter and its normal.

I hope the 6 will disappoint you less than it did to me. It’s the display mode in 3D view.
all sound familiar:

• wireframe
• solid
• material
• render

If wireframe and render are quite straightforward, solid and material are…
i’d say… more subtle

I use solid for texture view and never use the material view that runs on eevee, is slow and never gives me what i need as my scene is not setup for eevee.
My opinion ( though unimportant ) is that devs made some non user-friendly choices, but maybe you’ll show me am wrong

That is almost all you got to know for starting using blender.

hope you’ll like it^^

Happy blending !

[Mats_Ausland (Mats BA)]

Yes! Thank you so much for your reply! The video you sent me is exactly what I am doing! The guy in the video holding the Ipad is my supervisor! So thank you very much, I guess I will just have to get used to Blender at first! I got the S and G pretty much in the bag now! And changing fast through all the modes, using xray etc. I only wonder how i should attack the map, should i use it as a refrence, or should I just try to model out of the pictures i found online? Much like in the video.

[pitibonom]

I’d say, both

From my point of view, the best is to place the houses from the map you showed up in upper posts, drawing them in their overall shape and then add more details by looking at the photos.

It’s sometimes not easy to visually guess surfaces ratios from a photo as the perspective shortens what is far.
A smart tool that could insanely help you to unproject building faces is fspy.

Happy blending !

[RSEhlers]

As far as the Floor Plan I believe it was just a back photo of the ground plan…

Found another that is a bit better…

and comparing this to Map/world Views I don’t see any type of slant on the entrance wall…

@pitibonom Mentions Fspy in his post and I highly recommend using it…
Look at this…

Tutorial – How to use fSpy with Blender

I don’t know how much access you have for Photographs, but even Google Maps Street View will come in handy…

Since you want this on a smartphone you might look at this and it will show how to project the image onto your Building meshes and save tons of time in the process!

Blender 2.8 Texture extraction projection painting (part 1)