I'm making an Aztec forest scene, that will be go into my portfolio piece, any advice and critique on the following posts is much appreciated. i will updating the thread as I progress.
my workflow is focusing on the Walls, floors and ceiling. in this case the ceiling would be the trees. In which speedtree will come in handy. First of all is walls. I have created a Trim sheet and textured, although not finished, it needs feedback. I'm leaving it as is now, as I don't want to focus on too much as I can change it later on. Any feedback welcome I'm going to model the wall assets around this trim.
I have been cracking on with getting assets into this piece, utilising trim sheets and tiling textures to do most the work for me. I have mainly been focusing on getting the walls, floors and ceiling (trees) done in the scene so I can focus on unique assets and finesse.
to do : - vary the floor textures, sort out the tiling size as its a a little blurry for the rock texture. - speedtree assets needs a slight rework for what i am going for. - flowers, vines and rocks then onto other assets
any suggestions on how to make a water flow please let me know. I want to place a stream source in the little pond ( coming through the stone head) . I was going to look into panners and using a plane to bend around the model.
Any water shader advice would be highly appreciated
This is looking cool! I think you should plan out a colour scheme early on. Maybe drop the intensity of those light shafts.I feel like you could brighten the colours of the dirt and bark, will make texturing easier.
Thanks for your feedback I have been working on scene to get it finished. I've got about a week left till my deadline. here's my updated scene. I have been using a variant of tiling textures to get the bulk of the scene done and blended the textures with a moss overlay. using the same textures across the board to ensure the models blend well , however i feel like there may too much going on in the scene? or maybe I'm overthinking it. it is suppose to be a ruin and overgrown.
I still need to add some more vines attached to the models and finish the 3 statue sculpts. also a running water source needs to come through the head sculpt in the pond.
For the uniquely unwrapped models I have been creating a simple layered substance file i used for them all. this contained a base rock texture, with edges highlights of white, Overlayed with a moss dirt generator.
Using a mixture of sculpting and added normal stamps for the symbols, I think the result is nice. The archway is only sculpted asset here with a high poly bake, the other two are just low poly built. Here is the scene as of now. Lighting still needs to be tweaked once I get the candle light sources in the engine. Please any feedback is always taken on board and appreciated.
Need to sculpt some more assets and work on post process effects.
Added a new lighting pass and finished the assets, everything is now textured. Also added in some vines for the archway. Tweaked the lighting to gives more attention to the central path.
Added in some bloom scale in the post process effects just to brighten assets.
Factoring in some light shafts to add a nice effect to the scene.
I think you're knocking this out of the park man. I really like the lighting setup you went with. The only nitpicky lighting comment I have is the warmth of the light coming down the stairway might be too inconsistent with the directional. It does do a good job of highlighting the central path, but maybe a tweak to the directional or skybox could make it feel more like one light source. And this tree branch- seems a little strange to me. It could just be the angle but the shape looks strange. Looks like the trunk and branch are abruptly stopping and the transition into the foliage cards could be smoother. But that's all I can pick out. I've always wanted to do a Mesoamerican scene like this and this is pretty inspiring, good work!
Thanks for the feedback man, will tweak the lighting soon, i need to do a production build soon aswell so good timing haha. As for the branch its probably the model itself, I'm quite new to speedtree and getting the model optimized better probably has some thing to do with this. I will have to increase the length of the branch in future but for now i will just rotate the tree to a different angle i think.
Starting my planning with the Miro software. Using the Mood board approach of dotting my ideas, organizing myself with what needs to be done is a good strategy and stopped me in getting carried away and doing things on the fly e.g. creating without reference etc.
This allowed all my reference ideas and my planning to be in one spot and I could add to it constantly. I had 2 concepts to start with I wanted to take inspiration from, so I began to breakdown what I liked about them. It was mainly the composition and lighting of both concepts.
I took a lot of inspiration from the bottom piece e.g. the positioning of assets and the overall open composition. However I brought my own flare turned it into an Mesoamerican Scene.
As soon as I got the basics down in research, starting the hybrid approach was key. Getting the hybrid meshes done and experimenting with positions and overall composition cameras.
This was the very first progression shot on the scene.
Progressively adding in the hybrid meshes and furthering the composition.
I believe this was to be the final of the hybrid shots before I started working on foliage and other assets. After each shot I was given feedback which I worked on. to get an idea of where the foliage would be I ended up using green spheres as a substitute, this helped with getting ahead with the lighting direction.
Next I was cracking on with the production of the Trim sheet assets.
Production: - Trim Sheet Development So for the Trim sheet production I tried to follow the polycount tutorial on YouTube for some advice. I had a rough idea of what kind of trims i wanted. I was mainly using them for the walls and the back crypt structure. I created a 2k by 2k meter plane and divided it up into sections for what I needed. I wanted some ledge beams and some sort of stone brick sections with some symbols. Bringing this into Zbrush to sculpt some detail into the stone etc. I brought it back and forth into 3DS max and Zbrush to test the tiling etc. I ended up with a good result that eventually got a Moss variant textured over to break up the errors that were brought to my attention.
I sculpted some alphas maps into the design. However i realized i could just paint them on using a stamp later on, the result was mch nicer and visible when producing the normal map
I ended up taking the baked normal map into 3D Coat and stamping on the Alphas so i could get a curvature map with the added normal. Then bringing them into substance painter and using some tiling textures I created using alchemist to get the texture. Having gotten the trim sheet to a usable standard, I began to make some geometry out of it. I had an idea of walls and stairs however I got the crypt done first, unwrapping a blocked out version and manipulating the edge loops to make use of the trim sheet and add depth to the model.
It was a split between tiling textures and Trims for this piece.
As you can see I got the ledges and the overall look for the crypt done.
I was able to use the Trim sheet a lot throughout the scene which saved so much time later on. As i stated before, creating a moss variant to cover up tiling issues and other aspects of the trim sheet was very useful. Using a Trim sheet got my scene looking very well blended with nothing looking 'Out of Place'. The Rest of the scene was done with mainly Tiling textures and then unique assets.
In the engine I added a moss variant to the Trim sheet assets, I could vertex paint the meshes and add this wherever I needed.
- Low/High Poly structure
I have been following the usual practice of producing a low to high to low mesh build. Using a hybrid approach to the scene enabled me to get an idea of the shapes and types of assets I needed to make. Following reference I found, I produced an asset list consisting of textures, meshes and shaders I would need to complete the scene.
From this i began working on getting all modelled assets to a high polygon stage where needed. For example the archway, I had it ready to take into Zbrush and sculpted some edgewear for the worn down rock. This was basically shaving edges and making it look rough.
I took the high poly build and retopped the model in 3DS max. In baking this down into a normal map, i then took the model into 3D coat to stamp some added wear and tear using alpha materials.
Here is the Normal bake on top of a low poly build in substance. This is before the added Normals Here you can see the stamped normal from 3D coat taken in effect.
Here you can see the added normal maps and Tiling materials working together in substance.
This process was used for most of the unique assets. Only meshes such as the rubble were not sculpted and were only covered in a tiling texture, as it felt unnecessary to waste time on something that is a literally a pile of rubble and not really in the spotlight.
Using Alchemist to setup some nice basic tiling materials was really easy. I didn't use it for Landscape materials as the tiling was really an issue over a big scale however for smaller assets it is really great. Much nicer to use if you aren't that familiar with substance designer, in which I'm not.
Using basic untiled textures from the free website I was able to create some for myself to use in my scene all over. Also in substance painter for when in came to making smart material blends. The texturing process for most of the assets was quite simple in using the trim sheet material to cover them. With the texture done for everything.
Tiling assets were made to cover a lot of the models aswell including the rocks, floor platforms etc. All were varied in textures using vertex paint in engine aswell to stop repetition e.g. with a Moss overlay.
All tiling textures were made using basic textures found on 'Textures.com' https://www.textures.com/, I then broke them down and made them tile myself using Substance Alchemist which i found very useful in making quick and good looking textures.
I was able to make some textures with the software, I then took these to use in the engine, substance painter and speedtree for the trees.
- Unreal Shader Work Upon finishing the textures, i made some simple Master Shaders in unreal to help with organization and take some stress off of the engine when building.
First of all I made a Tiling Master material which used tessellation if needed. and with a vertex paint
This just makes it easier on myself and the engine in terms of time and efficiency. All of my tiling textures became an instance of this material Other Material shaders were slightly more complex to make, using tutorials I was able to make a material function to overlay a moss texture onto the rocks.
I was able to use this with tessellation to get my rocks looking quite nice with my base Rock textures and my own Moss material.
Here the Function being blended over the top of the rock texture.
Another shader I made following a tutorial from 'Game Dev Academy', I took this shader as it looked good for pond water, I added to it by overlaying a grimy dirt texture I made using Alchemist. using vertex paint aswell, I could add this effect wherever.
That's pretty much the shader setups I was using all around the scene.
- Presentation (Lighting, post process)
The final Aspects of the scene result in a lot lighting changes from night to day. The concept piece gave me inspiration for most of the lighting.
I wanted the fog effects and the central focus to the piece. However I also wanted there to be another direction to give a viewer some exploration of the scene, thus the side cave which is contrasted blue.
I ended up bringing everything closer together rather than being open to limit the viewers path throughout the scene. I did this by brining in a much more overgrown feel to the scene and contrasting the central with warmer lights to influence the eye.
By adding in bloom effects and volumetric fog it came out pretty good.
Reflection
Aspects I thought that went well consist of the overall process and quality of the work I produced compared to my previous projects of similar nature. I believe that i put more though into the planning and composition considerations when making this project thus benefitting me in the long run.
In one of my previous projects I was unfamiliar with using vital assets such as speedtree and trim sheets, having learnt new skills and getting better at prioritizing tasks I was able to produce most of the work within the first couple of weeks leaving more time for finesse. the project lasted over a a couple months.
In comparison to my work above, i have come on leaps and bounds, adding more colour variation and being able to produce more assets in the same time period thus creating a more detailed piece.
Comparing my work to an industry standard piece, the work on Naughty Dogs, Uncharted series can be found on Art station by a variety of high quality artists.
Using Pieces like this to compare against my own work.
From first comparisons, I am pretty proud off my work, the same feel for the overgrowth and quality of assets can be found in both pieces. However I feel my own is lacking the smaller details such as the higher grade sculpts and wall work. Everything is visible in the professional piece, mine however is slightly darker and you can struggle to see the architecture to its full potential. obviously I am comparing an vastly open scene with lots of daylight however I could made the scene slightly lighter so to capture more detail in the textures.
I could've added some more edges highlights on the trim Sheet to bring out the stone work. Especially on the walls below
Its a small issue but I feel it would added more visible detail to the scene. Especially in the smaller compact shots
In future I could add some more details to the scene post-deadline, for example some atmosphere effects using particles. Possibly to create a firefly scene.
Another add on would be the water source. I wanted to add a running water effect through the mouth of the statue head. It would makes sense to add a source to the water as it just looks flat.
Reflecting on the NLC, I feel like I have coped well with the requirements and followed the way I initially planned. Especially for this part of my portfolio piece. Taking in feedback well from lecturers aswell as external criticism, for example this polycount thread has received a number of replies to help guide me.
My second Project is the integration of Speedtree into my workflow. Using the software to produce the bulk of my foliage and just getting used to the industry standard Software altogether.
Planning:
My research regarding the scene began with getting to know the region i am designing the scene around. In this case it was Mesoamerican setting. The amazon rainforest was big influence in areas.
Using the Miro application i was able to plot reference and research clearly to help me out with what I wanted
So going for the thick rooted jungle idea with thick, chunky trees and spiral branches to give an overgrowth idea, for the plantlife is was bright colours and variation to fill the scene.
Production: Learning the Software was quite easy, there are plenty of basic tutorials to pick up quickly on.
This video by Adam lambert was a good head start on beginning to grow vines throughout the scene meshes. I used this to help to do so around the archway.
To start with the software I jumped right in and began creating some different trees. I have the textures made up for the bark already I started making some basic trees to get used to the software, for practice. Then moving onto my researched tree, the kapok. Taking inspiration from the thick lower base and spiralling branches at the top.
Furthering my design through feedback I ended up with this result.
Modelling other assets involved wrapping vines around certain objects. To create the thick roots i used a square box to act as a 'Attraction' to the branches within speedtree.
Using the 'forces' you can learn in speedtree and experimenting with the different ways the vines reacts to the mesh is quite useful.
Using these settings in the editor, I could customize the branch node and essentially grow the roots around the mesh. Instead of attraction you can also use obstruct or avoid, this is when the vines hit the mesh they can stop growing or just never meet the mesh. As I wanted the vines to wrap around the mesh, I chose attract. This meant as soon as the branch hit the mesh, it stuck to it.
As I progressed in learning the software, creating textures for the the branches was done though the use of textures.com. I found a branch texture and edited the colour. From the base colour image I produced an opacity map and a normal texture.
Using the hierarchy system shown (top left) you can plot the base, larger branches, all the way to the leaf meshes. You can even edit the geometry of the texture plane within the software so there's time saving instead of bringing your own geometry from 3ds Max.
As the models got more complex I noticed I got carried away with the detail, thus optimizing is quite key for using speedtree, i was noticing close up to 60k tris for my tree models. I learnt how to optimize the model by using the "Segments" section, you can control the level of loops in the model. This dropped the tris count quite a lot. I've seen some really detailed Scenes with models as low as 5k. I tried to hit this mark but ended up with my main tree model hitting 14k tris. A lot better than the 60 k the model was on before.
You can control the optimization through the widget here. from 0 -1 in percentage it controls the topology of the model.
Wind Control is a big aspect of using speedtree with unreal. I didn't mess around with the wind as much so i just setup basic adjustments to get the animation. Straight away its such a difference in using the 'SimpleGrassWind' . The branches are solid and towards the thinner branches you notice the wind take more action.
Here you can choose a type of vegetation to base the wind controls around. i chose palm tree for my model and edited the strength and direction in the settings to the left. When you export the complete model as a speedtree asset. the Wind Node in unreal takes all of the settings made and simulates them in unreal so you don't really have to anything once its in engine.
With this setting in unreal you can gain access to the same settings from speedtree, taking control of the speed and strength of the wind in the scene.
Shader aspects of the Speedtree were quite simple. it makes them for you on importing the assets. So grass, trees etc. get shaders and whatever material from speedtree applied to them straight away. The only tweak I made was to add the 'SpeedTreeColourVariation' node in the shader blueprint.
Adding this node allowed the colour to change on the tree based on its position, same for all foliage. Too much in the multiplier gave it a weird colour skew. however I set it on '0.1' which gave the colour a slight tint in variation. it helps break up the repetition in colour.
You can see the Difference in colour .
Reflection
I will be comparing my work to high quality vegetation that has been made using speedtree. Also previous work of mine to show how much it has improved.
I have made a demo scene in unreal showcasing the foliage I have made. Just as something to compare other work to.
To start I like how my vegetation turned out this time round, I think because I made more variation and used a proper software to create it, it turned out a lot nicer and cleaner in areas.
In comparison to other work i produced completely from scratch within 3ds max.
First of all this tree does look like a tree, i have produced some nice shots however resolution is too low and the branches are too consistent throughout, The bark texture is too largely scaled compared to my newer speedtree model, which is cleaner and more accurate to the tree type.
In terms of quality, using speedtree has made a huge difference from better textures and control over the modelling. Instead of modelling from scratch in 3ds max, you can control the shape aspects much easier and get a nicer looking tree.
In comparison to industry work, looking at marketplace packs you can buy which are going towards realism etc.
I think both have a lot of positives. To make my own work look better to this is probably the lighting method. Using distance field meshes instead lightmaps is quite new to me. In future i will continue to work with distance field meshes and improve on the lighting settings on my vegetation. I just think the shadows on sides of the foliage are to sharp and dark. Possibly looking into SSS maps to take into consideration the light shining through the leaf. the top work is very high quality, the resolution and detail in the flowers and the grass is high and vibrant. So one, look a the lighting setup in regards to shadows and work on creating a brighter field though the use of subsurface scattering is what I would do better.
Areas I could improve on with my tree model would be the abrupt stops to the mesh. I received some feedback on polycount regarding a model looking weird due to the fact the branch just cuts off.
In future work and use of speedtree, I will make sure my model looks correct. the reason this happened was due to optimization of the model. i had to reduce the length of the branches in order to stop the tris count getting higher and higher. this result in branches being cut and i didn't know how to overcome it. I have later realized but left too late now to fix for the deadline. I think in later tweaks to the project I will rework a new tree model with added knowledge of using the segments section in speedtree to cut down on triangle usage.
Replies
my workflow is focusing on the Walls, floors and ceiling. in this case the ceiling would be the trees. In which speedtree will come in handy.
First of all is walls. I have created a Trim sheet and textured, although not finished, it needs feedback.
I have been cracking on with getting assets into this piece, utilising trim sheets and tiling textures to do most the work for me. I have mainly been focusing on getting the walls, floors and ceiling (trees) done in the scene so I can focus on unique assets and finesse.
to do :
- vary the floor textures, sort out the tiling size as its a a little blurry for the rock texture.
- speedtree assets needs a slight rework for what i am going for.
- flowers, vines and rocks
then onto other assets
any suggestions on how to make a water flow please let me know. I want to place a stream source in the little pond ( coming through the stone head) . I was going to look into panners and using a plane to bend around the model.
Any water shader advice would be highly appreciated
I have been using a variant of tiling textures to get the bulk of the scene done and blended the textures with a moss overlay. using the same textures across the board to ensure the models blend well , however i feel like there may too much going on in the scene? or maybe I'm overthinking it. it is suppose to be a ruin and overgrown.
I still need to add some more vines attached to the models and finish the 3 statue sculpts. also a running water source needs to come through the head sculpt in the pond.
For the uniquely unwrapped models I have been creating a simple layered substance file i used for them all. this contained a base rock texture, with edges highlights of white, Overlayed with a moss dirt generator.
Using a mixture of sculpting and added normal stamps for the symbols, I think the result is nice. The archway is only sculpted asset here with a high poly bake, the other two are just low poly built.
Here is the scene as of now. Lighting still needs to be tweaked once I get the candle light sources in the engine. Please any feedback is always taken on board and appreciated.
Need to sculpt some more assets and work on post process effects.
Tweaked the lighting to gives more attention to the central path.
Added in some bloom scale in the post process effects just to brighten assets.
Factoring in some light shafts to add a nice effect to the scene.
Thank you though it is much appreciated
Planning:
- Miro
Starting my planning with the Miro software. Using the Mood board approach of dotting my ideas, organizing myself with what needs to be done is a good strategy and stopped me in getting carried away and doing things on the fly e.g. creating without reference etc.
This allowed all my reference ideas and my planning to be in one spot and I could add to it constantly.
I had 2 concepts to start with I wanted to take inspiration from, so I began to breakdown what I liked about them. It was mainly the composition and lighting of both concepts.
I took a lot of inspiration from the bottom piece e.g. the positioning of assets and the overall open composition. However I brought my own flare turned it into an Mesoamerican Scene.
As soon as I got the basics down in research, starting the hybrid approach was key. Getting the hybrid meshes done and experimenting with positions and overall composition cameras.
This was the very first progression shot on the scene.
Progressively adding in the hybrid meshes and furthering the composition.
I believe this was to be the final of the hybrid shots before I started working on foliage and other assets. After each shot I was given feedback which I worked on. to get an idea of where the foliage would be I ended up using green spheres as a substitute, this helped with getting ahead with the lighting direction.
Next I was cracking on with the production of the Trim sheet assets.
Production:
- Trim Sheet Development
So for the Trim sheet production I tried to follow the polycount tutorial on YouTube for some advice. I had a rough idea of what kind of trims i wanted. I was mainly using them for the walls and the back crypt structure. I created a 2k by 2k meter plane and divided it up into sections for what I needed. I wanted some ledge beams and some sort of stone brick sections with some symbols.
Bringing this into Zbrush to sculpt some detail into the stone etc. I brought it back and forth into 3DS max and Zbrush to test the tiling etc. I ended up with a good result that eventually got a Moss variant textured over to break up the errors that were brought to my attention.
I sculpted some alphas maps into the design. However i realized i could just paint them on using a stamp later on, the result was mch nicer and visible when producing the normal map
I ended up taking the baked normal map into 3D Coat and stamping on the Alphas so i could get a curvature map with the added normal. Then bringing them into substance painter and using some tiling textures I created using alchemist to get the texture.
Having gotten the trim sheet to a usable standard, I began to make some geometry out of it. I had an idea of walls and stairs however I got the crypt done first, unwrapping a blocked out version and manipulating the edge loops to make use of the trim sheet and add depth to the model.
It was a split between tiling textures and Trims for this piece.
As you can see I got the ledges and the overall look for the crypt done.
I was able to use the Trim sheet a lot throughout the scene which saved so much time later on. As i stated before, creating a moss variant to cover up tiling issues and other aspects of the trim sheet was very useful. Using a Trim sheet got my scene looking very well blended with nothing looking 'Out of Place'. The Rest of the scene was done with mainly Tiling textures and then unique assets.
In the engine I added a moss variant to the Trim sheet assets, I could vertex paint the meshes and add this wherever I needed.
- Low/High Poly structure
I have been following the usual practice of producing a low to high to low mesh build. Using a hybrid approach to the scene enabled me to get an idea of the shapes and types of assets I needed to make. Following reference I found, I produced an asset list consisting of textures, meshes and shaders I would need to complete the scene.
From this i began working on getting all modelled assets to a high polygon stage where needed. For example the archway, I had it ready to take into Zbrush and sculpted some edgewear for the worn down rock. This was basically shaving edges and making it look rough.
I took the high poly build and retopped the model in 3DS max. In baking this down into a normal map, i then took the model into 3D coat to stamp some added wear and tear using alpha materials.
Here is the Normal bake on top of a low poly build in substance. This is before the added Normals
Here you can see the stamped normal from 3D coat taken in effect.
Here you can see the added normal maps and Tiling materials working together in substance.
This process was used for most of the unique assets. Only meshes such as the rubble were not sculpted and were only covered in a tiling texture, as it felt unnecessary to waste time on something that is a literally a pile of rubble and not really in the spotlight.
Using Alchemist to setup some nice basic tiling materials was really easy. I didn't use it for Landscape materials as the tiling was really an issue over a big scale however for smaller assets it is really great. Much nicer to use if you aren't that familiar with substance designer, in which I'm not.
Using basic untiled textures from the free website I was able to create some for myself to use in my scene all over. Also in substance painter for when in came to making smart material blends.
The texturing process for most of the assets was quite simple in using the trim sheet material to cover them. With the texture done for everything.
Tiling assets were made to cover a lot of the models aswell including the rocks, floor platforms etc. All were varied in textures using vertex paint in engine aswell to stop repetition e.g. with a Moss overlay.
All tiling textures were made using basic textures found on 'Textures.com' https://www.textures.com/, I then broke them down and made them tile myself using Substance Alchemist which i found very useful in making quick and good looking textures.
I was able to make some textures with the software, I then took these to use in the engine, substance painter and speedtree for the trees.
- Unreal Shader Work
Upon finishing the textures, i made some simple Master Shaders in unreal to help with organization and take some stress off of the engine when building.
First of all I made a Tiling Master material which used tessellation if needed. and with a vertex paint
This just makes it easier on myself and the engine in terms of time and efficiency. All of my tiling textures became an instance of this material
Other Material shaders were slightly more complex to make, using tutorials I was able to make a material function to overlay a moss texture onto the rocks.
Following a tutorial by Lukas Ramblings on YouTube.
https://www.youtube.com/watch?time_continue=1862&v=HDluA1f5gzw&feature=emb_logo
I was able to use this with tessellation to get my rocks looking quite nice with my base Rock textures and my own Moss material.
Here the Function being blended over the top of the rock texture.
Another shader I made following a tutorial from 'Game Dev Academy', I took this shader as it looked good for pond water, I added to it by overlaying a grimy dirt texture I made using Alchemist. using vertex paint aswell, I could add this effect wherever.
That's pretty much the shader setups I was using all around the scene.
- Presentation (Lighting, post process)
The final Aspects of the scene result in a lot lighting changes from night to day. The concept piece gave me inspiration for most of the lighting.
I wanted the fog effects and the central focus to the piece. However I also wanted there to be another direction to give a viewer some exploration of the scene, thus the side cave which is contrasted blue.
I ended up bringing everything closer together rather than being open to limit the viewers path throughout the scene. I did this by brining in a much more overgrown feel to the scene and contrasting the central with warmer lights to influence the eye.
By adding in bloom effects and volumetric fog it came out pretty good.
Reflection
Aspects I thought that went well consist of the overall process and quality of the work I produced compared to my previous projects of similar nature. I believe that i put more though into the planning and composition considerations when making this project thus benefitting me in the long run.
In one of my previous projects I was unfamiliar with using vital assets such as speedtree and trim sheets, having learnt new skills and getting better at prioritizing tasks I was able to produce most of the work within the first couple of weeks leaving more time for finesse. the project lasted over a a couple months.
In comparison to my work above, i have come on leaps and bounds, adding more colour variation and being able to produce more assets in the same time period thus creating a more detailed piece.
Comparing my work to an industry standard piece, the work on Naughty Dogs, Uncharted series can be found on Art station by a variety of high quality artists.
Using Pieces like this to compare against my own work.
From first comparisons, I am pretty proud off my work, the same feel for the overgrowth and quality of assets can be found in both pieces. However I feel my own is lacking the smaller details such as the higher grade sculpts and wall work. Everything is visible in the professional piece, mine however is slightly darker and you can struggle to see the architecture to its full potential. obviously I am comparing an vastly open scene with lots of daylight however I could made the scene slightly lighter so to capture more detail in the textures.
I could've added some more edges highlights on the trim Sheet to bring out the stone work. Especially on the walls below
Its a small issue but I feel it would added more visible detail to the scene. Especially in the smaller compact shots
In future I could add some more details to the scene post-deadline, for example some atmosphere effects using particles. Possibly to create a firefly scene.
https://www.artstation.com/artwork/qAx06R Work by artist Julia Brasil did a nice piece with the same effects to add nice touch the final scene.
Another add on would be the water source. I wanted to add a running water effect through the mouth of the statue head. It would makes sense to add a source to the water as it just looks flat.
Reflecting on the NLC, I feel like I have coped well with the requirements and followed the way I initially planned. Especially for this part of my portfolio piece. Taking in feedback well from lecturers aswell as external criticism, for example this polycount thread has received a number of replies to help guide me.
My second Project is the integration of Speedtree into my workflow. Using the software to produce the bulk of my foliage and just getting used to the industry standard Software altogether.
Planning:
My research regarding the scene began with getting to know the region i am designing the scene around. In this case it was Mesoamerican setting. The amazon rainforest was big influence in areas.
Using the Miro application i was able to plot reference and research clearly to help me out with what I wanted
So going for the thick rooted jungle idea with thick, chunky trees and spiral branches to give an overgrowth idea, for the plantlife is was bright colours and variation to fill the scene.
Production:
Learning the Software was quite easy, there are plenty of basic tutorials to pick up quickly on.
Short videos I quickly watched were
https://www.youtube.com/watch?time_continue=30&v=Fc07a5ZcmP0&feature=emb_logo
https://www.youtube.com/watch?time_continue=3&v=V2bc93LGyp4&feature=emb_logo
This video by Adam lambert was a good head start on beginning to grow vines throughout the scene meshes. I used this to help to do so around the archway.
To start with the software I jumped right in and began creating some different trees. I have the textures made up for the bark already
I started making some basic trees to get used to the software, for practice.
Then moving onto my researched tree, the kapok. Taking inspiration from the thick lower base and spiralling branches at the top.
Furthering my design through feedback I ended up with this result.
Modelling other assets involved wrapping vines around certain objects. To create the thick roots i used a square box to act as a 'Attraction' to the branches within speedtree.
Using the 'forces' you can learn in speedtree and experimenting with the different ways the vines reacts to the mesh is quite useful.
Using these settings in the editor, I could customize the branch node and essentially grow the roots around the mesh. Instead of attraction you can also use obstruct or avoid, this is when the vines hit the mesh they can stop growing or just never meet the mesh. As I wanted the vines to wrap around the mesh, I chose attract. This meant as soon as the branch hit the mesh, it stuck to it.
As I progressed in learning the software, creating textures for the the branches was done though the use of textures.com. I found a branch texture and edited the colour. From the base colour image I produced an opacity map and a normal texture.
Using the hierarchy system shown (top left) you can plot the base, larger branches, all the way to the leaf meshes. You can even edit the geometry of the texture plane within the software so there's time saving instead of bringing your own geometry from 3ds Max.
As the models got more complex I noticed I got carried away with the detail, thus optimizing is quite key for using speedtree, i was noticing close up to 60k tris for my tree models.
I learnt how to optimize the model by using the "Segments" section, you can control the level of loops in the model. This dropped the tris count quite a lot. I've seen some really detailed Scenes with models as low as 5k. I tried to hit this mark but ended up with my main tree model hitting 14k tris. A lot better than the 60 k the model was on before.
You can control the optimization through the widget here. from 0 -1 in percentage it controls the topology of the model.
Wind Control is a big aspect of using speedtree with unreal. I didn't mess around with the wind as much so i just setup basic adjustments to get the animation. Straight away its such a difference in using the 'SimpleGrassWind' . The branches are solid and towards the thinner branches you notice the wind take more action.
Here you can choose a type of vegetation to base the wind controls around. i chose palm tree for my model and edited the strength and direction in the settings to the left. When you export the complete model as a speedtree asset. the Wind Node in unreal takes all of the settings made and simulates them in unreal so you don't really have to anything once its in engine.
With this setting in unreal you can gain access to the same settings from speedtree, taking control of the speed and strength of the wind in the scene.
Shader aspects of the Speedtree were quite simple. it makes them for you on importing the assets. So grass, trees etc. get shaders and whatever material from speedtree applied to them straight away. The only tweak I made was to add the 'SpeedTreeColourVariation' node in the shader blueprint.
Adding this node allowed the colour to change on the tree based on its position, same for all foliage. Too much in the multiplier gave it a weird colour skew. however I set it on '0.1' which gave the colour a slight tint in variation. it helps break up the repetition in colour.
You can see the Difference in colour .
Reflection
I will be comparing my work to high quality vegetation that has been made using speedtree. Also previous work of mine to show how much it has improved.
I have made a demo scene in unreal showcasing the foliage I have made. Just as something to compare other work to.
To start I like how my vegetation turned out this time round, I think because I made more variation and used a proper software to create it, it turned out a lot nicer and cleaner in areas.
In comparison to other work i produced completely from scratch within 3ds max.
First of all this tree does look like a tree, i have produced some nice shots however resolution is too low and the branches are too consistent throughout, The bark texture is too largely scaled compared to my newer speedtree model, which is cleaner and more accurate to the tree type.
In terms of quality, using speedtree has made a huge difference from better textures and control over the modelling. Instead of modelling from scratch in 3ds max, you can control the shape aspects much easier and get a nicer looking tree.
In comparison to industry work, looking at marketplace packs you can buy which are going towards realism etc.
Areas I could improve on with my tree model would be the abrupt stops to the mesh. I received some feedback on polycount regarding a model looking weird due to the fact the branch just cuts off.
In future work and use of speedtree, I will make sure my model looks correct. the reason this happened was due to optimization of the model. i had to reduce the length of the branches in order to stop the tris count getting higher and higher. this result in branches being cut and i didn't know how to overcome it. I have later realized but left too late now to fix for the deadline. I think in later tweaks to the project I will rework a new tree model with added knowledge of using the segments section in speedtree to cut down on triangle usage.
THIS IS FOR A MSC IN GAMES DESIGN