# cached BF

**URL:** <https://forums.chaos.com/t/cached-bf/57063>\
**Category:** General\
**Created:** [February 6, 2016, 12:33am UTC](https://forums.chaos.com/t/cached-bf/57063 "2016-02-06T00:33:33Z")\
**Posts on this page:** 10\
**Page:** 1

<div class="post-metadata">

**Author:** ![super\_gnu](https://avatars.discourse-cdn.com/v4/letter/s/cab0a1/32.png) [@super\_gnu](https://forums.chaos.com/u/super_gnu)\
**Post date:** [February 6, 2016, 12:33am UTC](https://forums.chaos.com/t/cached-bf/57063/1 "2016-02-06T00:33:33Z")

</div>

what with BF being so fast these days, the only reason i really depend on imap is for flythroughs, where “multiframe incremental” is a massive timesaver..

any chance we can get a “multiframe incremental” cached option for BF? or is it just impractical? i guess it would mean either modifying imap to do an unfiltered per-pixel solution (currently doesnt work, even with interpolation at 0 (or 1 i dont recall) and min and max at 0, it produces a horrible result much slower than BF) or separating the BF into a prepass, which might be rather a big job.

---

<div class="post-metadata">

**Author:** ![vlado](https://avatars.discourse-cdn.com/v4/letter/v/779978/32.png) [@vlado](https://forums.chaos.com/u/vlado)\
**Post date:** [February 6, 2016, 1:13am UTC](https://forums.chaos.com/t/cached-bf/57063/2 "2016-02-06T01:13:04Z")

</div>

It might be useful to think about it…

Best regards,  
Vlado

---

<div class="post-metadata">

**Author:** ![Morbid\_Angel](https://dub1.discourse-cdn.com/chaosprd/user_avatar/forums.chaos.com/morbid_angel/32/21272_2.png) [@Morbid\_Angel](https://forums.chaos.com/u/Morbid_Angel)\
**Post date:** [February 7, 2016, 2:31am UTC](https://forums.chaos.com/t/cached-bf/57063/3 "2016-02-07T02:31:52Z")

</div>

in fact, I’m doing a job right now where its simple geo in the exterior setting. I cached irmap/lc but there were splotches so I switch to bf/lc. Render time went down from 50 min per frame to 30!

---

<div class="post-metadata">

**Author:** ![super\_gnu](https://avatars.discourse-cdn.com/v4/letter/s/cab0a1/32.png) [@super\_gnu](https://forums.chaos.com/u/super_gnu)\
**Post date:** [February 7, 2016, 7:24am UTC](https://forums.chaos.com/t/cached-bf/57063/4 "2016-02-07T07:24:19Z")

</div>

yep.. as ive noted before.. the new BF is great.. but the imap needs some love 🙂

---

<div class="post-metadata">

**Author:** ![vlado](https://avatars.discourse-cdn.com/v4/letter/v/779978/32.png) [@vlado](https://forums.chaos.com/u/vlado)\
**Post date:** [February 8, 2016, 9:21am UTC](https://forums.chaos.com/t/cached-bf/57063/5 "2016-02-08T09:21:25Z")

</div>

> [@Motion Blur and animated vray proxies](https://forums.chaos.com/t/motion-blur-and-animated-vray-proxies/44204/17):
>
> in fact, I’m doing a job right now where its simple geo in the exterior setting. I cached irmap/lc but there were splotches so I switch to bf/lc. Render time went down from 50 min per frame to 30!

I’m working on some further optimizations that would bring about 20-25% speed increase for the BF+LC combo…

Best regards,  
Vlado

---

<div class="post-metadata">

**Author:** ![super\_gnu](https://avatars.discourse-cdn.com/v4/letter/s/cab0a1/32.png) [@super\_gnu](https://forums.chaos.com/u/super_gnu)\
**Post date:** [February 8, 2016, 9:40am UTC](https://forums.chaos.com/t/cached-bf/57063/6 "2016-02-08T09:40:16Z")

</div>

wow.. as always.. nice to hear…! still be good to have a caching option though 🙂

---

<div class="post-metadata">

**Author:** ![Dave\_Wortley](https://avatars.discourse-cdn.com/v4/letter/d/ecae2f/32.png) [@Dave\_Wortley](https://forums.chaos.com/u/Dave_Wortley)\
**Post date:** [February 8, 2016, 10:55am UTC](https://forums.chaos.com/t/cached-bf/57063/7 "2016-02-08T10:55:46Z")

</div>

We’ve just had a bit of a nightmare job which was a 5000 frame long sequence. It was too slow with BF/LC and as it was just a camera move, no animated objects we did IR + LC with a pre-baking pass first, the dynamiscism of the camera move meant we had to prepass in some cases every 2nd frame rather than every 10th… We ended up breaking the sequence into sections, rendering these separately so we use more machines to cache the prepass. We still had loads of flickering issues in the fine detailed areas. It was a horrible scene, very low gloss flat colour surfaces with fine bump maps, paper-thin modelling details.

For camera fly-throughs, we could really do with a more advanced caching system now, being able to bake the scene with BF could have really helped!

---

<div class="post-metadata">

**Author:** ![vlado](https://avatars.discourse-cdn.com/v4/letter/v/779978/32.png) [@vlado](https://forums.chaos.com/u/vlado)\
**Post date:** [February 8, 2016, 11:06am UTC](https://forums.chaos.com/t/cached-bf/57063/8 "2016-02-08T11:06:29Z")

</div>

If you think you can send me some part of that scene for investigation, it will be very helpful. From the description, it looks like the GI was not the real problem, but it would be good to take a look anyways.

Best regards,  
Vlado

---

<div class="post-metadata">

**Author:** ![Dave\_Wortley](https://avatars.discourse-cdn.com/v4/letter/d/ecae2f/32.png) [@Dave\_Wortley](https://forums.chaos.com/u/Dave_Wortley)\
**Post date:** [February 8, 2016, 11:09am UTC](https://forums.chaos.com/t/cached-bf/57063/9 "2016-02-08T11:09:51Z")

</div>

> [@SSS2 Artefacts/flickering](https://forums.chaos.com/t/sss2-artefacts-flickering/49965/6):
>
> If you think you can send me some part of that scene for investigation, it will be very helpful. From the description, it looks like the GI was not the real problem, but it would be good to take a look anyways.
> 
> Best regards,  
> Vlado

With pleasure, I have already zipped it up for you!

---

<div class="post-metadata">

**Author:** ![3LP](https://avatars.discourse-cdn.com/v4/letter/3/5f8ce5/32.png) [@3LP](https://forums.chaos.com/u/3LP)\
**Post date:** [February 8, 2016, 2:37pm UTC](https://forums.chaos.com/t/cached-bf/57063/10 "2016-02-08T14:37:31Z")

</div>

> [@On another topic](https://forums.chaos.com/t/49964/13):
>
> I’m working on some further optimizations that would bring about 20-25% speed increase for the BF+LC combo…
> 
> Best regards,  
> Vlado

Nice! If GPU can help for the task, even better 🙂  
Actually, could we have the BF cache for RT GPU as well?

Stan
