Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - mase

Pages: [1] 2 3 ... 9
1
I might have agreed to post on a forum and have others see my posts - that's about it, not more than that.
So, sure, go on and have a blast. Just make sure my posts aren't included in your 'research'.

That's not really my goal here, and if I'm in wrong here with this concept then I won't proceed with the idea. Besides, I don't even know how one would implement it and was more interested in discussing a tool that would be useful not only for me but for all the community. If it's not within forum rules then I guess there's no reason to continue this discussion.

2
All the obvious (and questionable) use cases aside - have users ever agreed to their posts (and potential solutions and work results) being scraped off the forum like that? Don't think so... and honestly, wtf?
Posting to a forum and having it visible to forum members is different from having someone scraping it for 'content gain'. Do the leg work and learn on your own just like everybody else.

Well, in a way there's no one stopping you from just going through each forum thread, right-clicking and pressing Save Page As... I believe by posting something on this forum you also agree to have that information public. Perhaps using this information to then gain something monetary out of it or distributing in some way could be questionable but other than that I don't see how archiving public information crosses any ethical boundaries (if that's what you're trying to imply).

Essentially, I see this Google chatbot as a more advanced search engine that is also able to connect various findings between the sources it's been provided and summarize it. Similar tools have been on the market forever.

3
That would be nice. Not sure if NotebookLM allows to integrate its' chatbot elsewhere but I would find this feature very valuable (and I believe some others too).

We could also avoid having duplicate threads for the same topic (because forum search, or even google search for that matter doesn't always find other threads on the same topic). Another very useful feature that Notebook LM chatbot has is it references sources after each statement so one can go and check it themselves. I imagine this would also help when looking for a particular thread or post and not remembering the exact title or content of the message (I was looking for an old Juraj's project post sometime ago and even fully knowing what it looks like and vaguely remembering the title could barely find it).

Besides that, I'd say forum works pretty well as it is so I'm interested to hear what are the main reasons to update it (if not for an easier/better search).


4
I see. What about other information on this forum? People discuss in all kinds of topics related to rendering, hardware, workflows etc. I, for example, have a few threads bookmarked for when I need to come back to certain topics and look up information because I know it's credible and it works. We could basically have a forum chatbot, and of course as I mentioned previously not all of the information here is credible or still valid but that's for everyone to decide I guess.

5
I really like the lighting, especially in the first image. I've noticed you've used it in your past projects too, and it really has a great contrast and white balance to give off this photographic look. Mind sharing more details about the lighting?

Great work, keep it up.

6
So recently Google has posted a research tool called NotebookLM, and to put it simply it's a custom AI bot (similar to Chat GPT) that only looks up and answers questions found in data one provides. Unlike traditional AI bots, it doesn't hallucinate information if it doesn't know it and unlike other bots can answer you 'I don't know' which can be very useful.

People started to use it to make custom AI bots that are very niche and topic-specific which basically act as a big manual for your chosen thing, except that it has AI that's quite efficient at this. So my thought was - would it be possible to download certain threads of this forum, put it into a PDF and use it as a source? There's a lot of valuable information here (and a lot of not so valuable, opinions, out-dated information etc. but still) and it can get troublesome when sometimes you need to find a specific topic or question.

I tried making a quick python script that fetches information from threads but it's not as easy as I thought and perhaps there's an easier solution that I don't know. I'm also curious if anyone else is using the tool or finds this idea interesting and perhaps relevant for them.

7

And I guess to relate to the thread, I would also assume that most studios / artists who mainly produce archviz don't need to mess with color management settings besides simply keeping their monitors calibrated and using those profiles correctly.

it's not about the subject but the delivery technology: most phones nowadays have P3 screens. this means that non colour managed Corona UNDERSTANDS only 67% of colours a small device in your pocket is able to DISPLAY. that does not sound very PBR, does it?

No it does not and perhaps I assumed wrong.

Please share your opinion and expertise on this matter - since we're living in times where media is consumed on all kinds of screens - should archviz studios bother with color managing their workflow? Again, my knowledge is very limited on this topic but I'm always eager to learn.

8
Interested to hear what other TM setups are you using (since you've mentioned that slapping AGX TM is only necessary for certain cameras)?

I'm also a big fan of 2 LUT AgX tonemapping setup that was shared in this forum, and I believe combined with Tone Curve it pretty much works in any kind of scenario (at least visually speaking). I also hear that some people use ACES / AgX interchangeably, depending on lighting situation, so I'm always very eager to know what other TM setups are people using out there (of course, if it's not a company secret or something).

And I guess to relate to the thread, I would also assume that most studios / artists who mainly produce archviz don't need to mess with color management settings besides simply keeping their monitors calibrated and using those profiles correctly.

9
It seems that Vantage allows 2 GPUs but there are some drawbacks which you should look into (there's an older thread on Chaos forums: https://forums.chaos.com/forum/chaos-vantage/chaos-vantage-general/1171617-chaos-vantage-multiple-rtx-gpu-different-series)

On the other hand, 7970x is a powerful CPU and I assume you will be pairing it with quite a bit of RAM too which indicates you will be working on heavier scenes. So, if you have a large scene most likely you will run out of VRAM regardless whether you have a single or dual GPU setup (from personal experience, it is common for large and detailed scenes to "eat" 100GB of RAM or more). On the other hand, if you optimize your scenes to use let's say 20-30GB of RAM then you also might as well render them traditionally using Corona and all of those 32cores on 7970x. Remember that Vantage still lacks certain functions that Corona has and there's also going to be various bugs that will have to be dealt with.

I'm still not sure what's the best case project to render in Vantage if hardware specs are of no issue. Perhaps small interior scenes where you need many animations in one tight scene? Because let's be real, who owns dual GPUs with 20GB of RAM or more on each to validate GPU rendering. I'd love to hear other opinions on this too.

10
In our studio, we have FP preset library with custom made presets for grass, bushes, trees, people etc. So moving to Chaos Scatter would be a hassle.

On the other hand, 90% of the time we only use the basic functions of FP and on very rare occasions do we turn to effects or other specific parts of FP. So in reality, most of it could definitely be achieved in Chaos Scatter. But a few times that I have tried I missed a few key functions from FP and so I switched back again.

Interesting to hear what others think.

11
Thanks, that's valuable info. And yes, that's just me being lazy and not testing it out myself. Perhaps a quick test scene and a few comparison renders would yield more precise results. But otherwise it's great to know how OOC works under the hood.

12
So just to be sure - if I have 8k texture in my scene but the object in camera appears far away, OOC will create let's say 256x256 image version and will discard anything above that? Also, do we know how much RAM does this conversion process eat? As in, if the texture is 1k or 8k and it needs to be downscaled to 256x256. Does it impact render times in any significant amount?

13
I guess all of these suggestions more or less come from personal experience and studio setups but in our office we have render nodes with 200gb of RAM and even that sometimes is not enough. Even if it's not an extremely large project, it is easy to lose track of assets and textures you have in your scene and you don't want to spend time optimizing your scene after you've built everything the way it should be.

I'd say aim for max amount of RAM that your wallet can carry. Even if you don't need it now, you will in the near future, as all the assets and everything are getting more and more detailed, textures are getting larger and overall the need for realism calls for complex and elaborate scenes.

And this also translates to PCI Gen 5.0. You might not need it at this exact moment but it's definitely going to be needed in a few years, given how large the files are getting and how fast we want to access them. Last thing you want is to build a new pc and in a few years start to feel it bottlenecking in your workflow (and all of this is valid only if you can afford it of course).

14
So this might be answered somewhere else but I always wondered whether Corona out-of-core textures works the same as manually resized textures, in terms of RAM performance? Is there any benefit to resizing textures in your scene to smaller resolution, or does this Corona option make it useless (or at least the difference is so marginal that it's simply not recommended)? Really curious whether someone has tested it out.

15
Hardware / Re: GPU - 3ds max
« on: 2025-03-12, 11:02:12 »
A little late to the party but could you elaborate on the 256GB RAM that you chose for your setup? I'm curious about the specific set you got, and perhaps an update later on how it performs.

Pages: [1] 2 3 ... 9