OctaneRender® 1.024 beta 2.48b TEST (lin/win) [OBSOLETE]

[Jaberwocky]

I can confirm that enlarging the render preview window to full screen now automatically sets the image to full size as well. Nice one. Thanks Marcus.

[funk]

Scratch that... user error

The autofocus and 3d stereo render buttons are missing... not that I use them, but I noticed they are gone

[SamCameron]

[Jaberwocky]

Another query.This time more of a logical one than a bug.
In a real camera analogy a depth of field is created by the lens aperture.
F1.4 is very fast letting in lots of light but with a very shallow depth of field.
when you get down to around F22 you have almost infinite depth of field but requiring a longer exposure to get lots of light to get an image.

Surely these two controls I have highlighted in the attachment should really be set as one slider only, if the Camera lens analogy is going to work.This slider should really be labelled as a series of F stops from say F1.2 to F22.

You already have the autofocus point eye dropper tool on the tool bar to focus the lens on the relevant point in the scene , as an autofocus lens would in real life.Couple that with the aperture slider described above. That should duplicate a Camera lens correctly.The remainder of the camera controls you already have.The ISO Slider should be re label'd to ISO settings 64 / 100 / 200/ 400 / 800 / 1600 / 3200 / 6400 as they are currently on most Digital SLR's.The scene image would get progressively grainier as the slider approach's the higher ISO values. This way anyone who uses octane and who also does photography should instantly be able to replicate a photographic situation.

The current interface sort of smacks of over complication.

Any other views on this suggestion ?

[Jaberwocky]

Sorry , forgot the image....Ooops !

[Jaberwocky]

Sam

This now accurately reflects the amount of available memory you have left to load a scene into Octane after your video card has powered your display/s.This of course would go up to the max available memory on board if you were running Octane on a spare card.Typically you loose somewhere between 100 mb and up to 3-400 mb of memory depending on how many screens you are running and what resolution and bit depth they are.

[Ayertosco]

Your point made me dream about lens\camera profiles. Being able to shoot-render with a 35/f2.0 Leica Summicron would be sweet.

[Jaberwocky]

Your point made me dream about lens\camera profiles. Being able to shoot-render with a 35/f2.0 Leica Summicron would be sweet.

Yes and a hell of a lot cheaper than a new Leica M9

http://www.reddotcameras.co.uk/about_us.php

[Ayertosco]

Your point made me dream about lens\camera profiles. Being able to shoot-render with a 35/f2.0 Leica Summicron would be sweet.

Yes and a hell of a lot cheaper than a new Leica M9

http://www.reddotcameras.co.uk/about_us.php

Uahahah, not spending 10k $ is of course one of the big advantages

[ROUBAL]

Hi, the storm has gone away now, and I have been able to do some more tests with 2.48 and 2.48b as well. I reported the results of my experiments in the 2.48 topic, but I give them again here because some of them are the sames as 2.48b results :

I have taken my old good Micronta stopwatch and compared 2.47 and 2.48/2.48b behaviour :

First, I have used a scene of average weight, with 4 detailed cars, a phone booth and some other props, for a total of 4 millions triangles.
The memory amount used is 1414.3/2880MB (in 8192x4096).

I first started with the scene displayed and rendered in 1280x720, and the navigation is fluent in both 2.47 and 2.48.
Then I switched to 8192x4096. Obviously, the navigation is no more fluent, in both versions, but the big difference is noticeable when using the material picker to select materials :

In 2.47, when you click on an object in the render view, the picker cursor changes instantly into an arrow and the materials panel shows up instantly, and when you adjust a parameter with a slider, it is modified fast in the render view. The time seems dependent on the object selected, and vary from 4 seconds up to a bit less than 10 seconds in my test scene.

In 2.48 (2.48b), when you click on an object in the render view, the
picker cursor remains as it is for 4-5 seconds before changing to an arrow, and beeing usable to click on a slider in the materials panel.
if the materials panel is not yet open, it doesn't show up until 4-5 seconds. Ialso noticed that it is view dependent : when clicking on an object seen in close up, the time before getting the arrow cursor and material panel has sometimes increased up to 40 seconds (measured in 2.48b), which is huge and really not good at all for the workflow !

When I modify a parameter by moving a slider in the materials panel,
it takes 14, up to 20 seconds before the material modifications are applied in the render view. This seems also view dependent, or the duration variations can also be dependent on the instant of the click in the current rendered frame ?

So, 2.48/2.48b are actually very slower and less interactive than previous versions.

Now, the good news :

1 - The alpha transparency works again.

2 - I can see in 2.48/2.48b that the Zoom/Framing of the static image by Ctrl-LMB is much faster.

I also noticed that the direction of the static zoom is inverted in this version, requiring a modification in the preferences to behave like in 2.47, but as the new direction looks more logical for me, I don't consider this as a bug, but more as a bug fix!

3 - And the best news for the end : I loaded my very heavy London City scene, with 8.7 millions of triangles, and set it to 8192x4096.
Then I relinked by using the "Load new file" button a heavier scene with one more building. This scene contains 9.86 millions of triangles and feeds my memory with 2609.9/2880 MB.

I have 8GB of physical RAM, and Win7x64 resource monitor shown off more than 15GB required. It took more than 20 minutes to relink using the swap file, but it relinked well ! So it is a good point for people (like me) with low end motherboards not supporting much RAM : The relink process is very slow, but it works ! Give an eye at the proof screen below.

My temporary conclusion : this release shows a great improvement, but requires some work on fixing abnormal high delays when using the materials picker, and a faster refresh system.