Visualization and the visual language as a whole is a crucial phenomena when we communicate our ideas to the clients. Should that be vectorial, artistic, non-realistic, photo-realistic, 2D or 3D from classic quick shaded representations to high-end VR or XR, most of us would agree that the way we communicate our ideas is truly essential.
I would like to start a dedicated conversation and hear your feedback on this topic. Please let me know:
Let me know if you have anything else in connection to this topic. Let's start the conversation.
Improve the sketch render by tapping into leading edge AI tech like Stable Diffusion, GPT-3. Should be able to feed it the Archicad 3D window and receive a convincing hand drawn sketch generationally ahead of how sketch lines work in Sketchup. Apply this to all drawing outputs in Archicad for very early sketch design stage.
- What type of visualization do you do? (Quick Shaded, Rendering, Animations, Real-time visualization, ...)
Pretty much all types - however within Archicad we typically stick with the shaded or white model renders.
- What is your current workflow for visualization? (In-house, Outsourced, Partially Outsourced, ...)
Varies widely depending on the type and stage of project, as well as the time constraints and availability of skilled operators in-house.
- What are the main challenges or obstacles in your current workflow?
Good quality but low poly landscape and manufacturer content for Archicad, as well as missing export formats. We also often use BIMx, thus it would be great if I had better rendering abilities.
- What third party solutions do you incorporate in your workflow if you do any?
VRAY, TWINMOTION, Keyshot, Photoshop & Illustrator
Generally we agree with some of the other comments above, that Graphisoft should focus on Archicad's core functionality (do not know anyone who bought Archicad for its rendering abilities!), best in class export formats and API's, rather than switching render engines every couple of years.
Hi! First of all, thanks Graphisoft for starting to engage the community more actively. As a suggestion on that, I think, we all would like somehow to see how all these ideas and suggestions that the community is giving, are translated onto some kind of roadmap or even some intention, just to understand what has been selected or understood from all the comments, etc.
Secondly, replying to your questions:
We never use Archicad render engine (C4D or redshift) to do renderings. We did in the past, but it was too complicated and too time-consuming for a mediocre result. Only "experts" in the office will be able to create a nice rendering and it will take a really long time, not only rendering-wise but trial and error to get the settings right. Also as other fellow community members have already said, Archicad doesn't have the right object libraries, and trees, and isn't even able to handle high polygons or large scenes with many instances.
Currently, we have 2 workflows for our renderings and images:
We use an external renderer, mainly twinmotion, but sometimes (rarely) Unreal or Blender, for the High quality renders. These renders usually need to be very good and/or very fast and always need very high-quality assets, especially vegetation.
For everyday images, work-in-progress, coordination, or even presentation of a more diagrammatic style, we use either 3D documents or more and more, screenshots from the 3D window. Like @Tim Ball mentioned earlier, this is a number one priority in most of the people's workflows, being able to use HIGH QUALITY 3D captures from the model in the views and layouts. At the moment the quality that the views provide is so bad that everyone I know does screenshots and then bring them as an image... quite a terrible workaround, as when you need to update your image after a model change, you need to repeat the whole manual process and almost never the image looks the same (ratio, camera, etc...) So this should be a priority to enable quick and easy images that look good and are updated in real-time with the model changes.
Following this, I think the rendered sections and elevations are a really nice addition to the conceptual design workflow, but there are 2 major flaws that make them almost unusable:
1. This rendering style should be a Model View Option and not part of the section settings. Now, we need to duplicate sections in order to carry both sets of documentation: conceptual style AND standard technical documentation. Using an MVO will allow using a single section for as many styles are possible.
2. This workflow doesn't work for plan views, which is very annoying. We need to use Graphic overrides for this and create dozens of overrides for the different materials. This mess up our files as all of these overrides rules make the search for other rules more difficult.
As mentioned in other comments and threads, shadows could be really revamped: being able to show real shadows, even when the model has a cutting plane or similar will improve a lot the model visuals and the rendered documentation.
Better handling of instances will definitely help also to populate our models with trees, people, and other assets.
Also, instead of creating every year a new additional image catalog that is somehow detached from the template and that won't satisfy everyone, GS could add a procedural surface creator, that creates also the corresponding surface cover fill. This is not science fiction, there is already one online, that of course has only link to revit...
This would enable almost infinite access to surfaces that are also seamless.
Rendering using Cinerender and Sketchrender engine.
In house, using Cinerender and Sketch render engine.
Lack of 3D presentation library parts that are detailed enough, like trees, people. Currently we work around this by keeping the style more 'hand drawn' than photorealistic. This our main workflow, since we prefer output that is ready in a few minutes. However if we do want to scale up to a sales brochure higher level, the library is lacking.
Photoshop for postproduction and combining the cinerender output with sketchrenders. We overlay several renders: a colour render as base, a render pass that renders ambient occlussion in a black and white pass, a render that renders trees and other foliage as a black and white contour in order to be able to remove their contours from the sketch render and several sketch renders for outlines, vectorial hatching and interior lines.
The out of the box cinerender scenes are severly lacking. The 'fast' render is slightly too low quality and anything medium and up takes a rediculous amount of time. I took the effort to delve in to the advanced settings and change them as needed to get a workflow that at the time of AC18 release could 'rival' with Artlantis Render in terms of output quality and render time. I can imagine that anyone new to Cinerender will get put off by the quality or the render times, while it certainly can work well enough. I'm sure that by now there are better 3rd party options, but we are content with what ships 'for free' with Archicad.
Thank you for all of your kind engagements in this topic. Let me roll the discussion further with another question. Many of you already gave an insight into this, but I am curious to hear more about the following:
Do you post-process your visualization? What is your workflow and what software are you using for that?
Very little post processing is done after I get an image from Enscape. When I make any amendments it is done in Adobe Photoshop where I will enhance the levels, highlights or shadows a little bit. Other than that it is straight out of Enscape to the client.
not too much post production after the output of a Twinmotion path tracing image generally I switch to photoshop and I use the camera's RAW filter to adjust the contrasts and modify the colors, generally the adjustment that is used in post-production is Level, contrast, brightness, exposure, gamma, and color balance filter in post production, in short if you want to give us post production options, just give access to the effect options of post-production of cinerender after rendering is finished
I think some of these options are there, but I found that it increases render time from my 'fast' settings and it's a simple adjustment layer in photoshop to reduce saturation and fix brightness etc a bit.
This is the same reason we do a 2nd seperate AO render. It is faster to render a colour render without AO and the black and white simplified AO and just multiply them together in photoshop.
We need photoshop regardless of our artist impression for other workflows too, so I just noticed that rather than endlessly doing test renders and getting all the settings just right for each project, it is quicker to use an adjustment layer in photoshop.
These type of renders should not take more than 2-3 minutes for 195x135 mm at 300 dpi for us, since we will make these every time there are design changes to communicate with the client. So efficiency is key.
I don't really see anyway to implement this in cinerender other than having it 'baked in' with your image as it is now and I find that to be a slow process.