Hi everyone!
I’d like to share the exterior views of an extensive renovation project involving both the front elevation and, more significantly, the rear of the property, where the living area has been extended with a rear extension, while the bedrooms have been enlarged through three new volumes connected by a continuous roof.
The project is located in the UK, where metal roofing is quite common, along with the use of natural materials such as timber cladding.
For this project, I was involved in both the architectural design, working on behalf of a UK practice, and the architectural visualisation, producing all of the renders myself.
I’m currently putting the finishing touches to the interior views of the living area, which I’ll be sharing once they’re complete.
Software: Allplan, Cinema 4D, Corona Renderer Post-production: Photoshop + Magnific AI for upscaling
Great set of images. It’d be interesting, as always, to see the raw versions before AI, which here seems to work really well, other than the woman’s fingers and elbow in the interior shot, which is easily fixed and is a very cool shot nevertheless
Hi! Thanks! I’ll include some of the views in their raw version, exactly as they came out of Corona. I use AI to add people (not always, sometimes I use it to improve the 3D people models) and for upscaling to achieve some extra detail.
Our forum will be improved so that questions like this wouldn’t be necessary, but meanwhile, it would be good to share some additional details, at least when posting in the Common category.
Did you use this ID mask to directly help AI or to do some compositing between raw renders and AI ones?
I mean..for example… you ask AI to enhance the raw render..then you take the output, you run Photoshop and you put it over the raw render and then you use the ID mask to choose what to keep or discard from the AI.
Anyway…it’s clear that you base renders are alread y very detailed and really well done.
It’s a pity that, if you use an AI online tool, you’re feeding and training its database and, in a near future, your skills to get something like you get out from Corona will be available for anyone, indipendently from his knowledge and expertise…but it seems that the mood is “who cares” so…this seems the trajectory for this industry…so let’s see what will happen…
Hi Bardo, no, I normally open the raw render in Photoshop and make the usual adjustments in Camera Raw (contrast, white balance and vibrance) I then resize the image so that the shorter side is no more than 2000 px. This is because Magnific uses fewer credits that way, while the starting image is still large enough.
I then upscale it without adding any prompt or other instructions. After that, I bring it back into Photoshop, place it over the original render and use a mask to decide which parts of the upscaled image to keep, and whether to use them at 100%, 50%, 25%, and so on. I do not use the ID masks for this, I simply judge it by eye based on the image.
I only created these ID masks because I was asked to show the clay renders, although I think the process is easier to understand with the ID masks included.
As for AI training, I have reached a sort of “who cares?” philosophy. It is a tool that needs to be regulated, and will need further regulation, but regardless of what I personally choose to do, it will continue to develop with or without me. Unfortunately, I do not have the time to learn ComfyUI or other locally run tools, so for the time being I have to rely on cloud based solutions.
Process is clear… I was thinking that you could use the ID mask as precise masking way… anyway it’s clear your workflow.
About AI, please, don’t take it a personal attack… mine was just a general thought about AI, inspired by the quality of your out of Corona images…(but it’s the same for any kind of out of renderer high quality iimage).
No, no, don’t worry at all I absolutely didn’t take it as a personal attack. These are discussions I often have with other people in our profession, and we are all questioning what the future will look like and the impact AI is having, and will continue to have, on our work. Personally, and of course I can only speak for myself, as I know others are experiencing the exact opposite, over the past couple of years I have seen a considerable increase in the amount of work I receive. I don’t know whether this is partly because AI can help speed up workflows and make it possible to provide content that would previously have been unthinkable without excessive costs and production times, particularly animations. However, despite the speed AI offers, and the fact that it allows you to improve and add apparent “quality” to images that are only roughly developed, I still prefer to start with final images that are already carefully resolved and then use AI to add that extra layer. At the moment, you still do not have complete control over the result or full consistency between images, and for certain types of work that is absolutely essential.
Query this: Has rendering already become commoditized? Or is it still in the process of becoming a commodity?
Long story short:
If there’s a historical rhyme, it’s drafting. CAD commoditized the drawing, but it didn’t commoditize the architect — it just eliminated the people whose entire value was the drawing. Rendering is about two-thirds of the way through the same curve. (per claude reasoning)
There will always be the boutique 1% that rely on the traditional service but it’s already become a commodity and will continue down that path.
I partly agree. Draughtspeople still exist today, and there are plenty of people who do not design but simply carry out the production work by drawing. They use CAD rather than drawing by hand, of course, but the tool has changed, not the purpose of the role. I am not sure whether the same will happen with AI. The real risk, in my view, is that the decision making process itself will be handed over to AI, and that is the most worrying aspect. Personally, that is not something I want to do. I want to retain control and decide how, what and where.
