<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class=""><div class="">Hi Neil, thanks for your thoughts!</div><div class=""><br class=""></div><div class="">I thought I’d give you some answers now, although I’m still working on trying to understand what’s going on with the particular dataset that gave me the incentive to write my original question.</div><div class=""><br class=""></div><div class="">-as for judging the overlap in the BDV, I think that one can get a pretty accurate idea of the registration by zooming in on single beads. Subjective, yes, bit if a registration is clearly off that’s normally easy to tell. What I find very difficult to understand is why some angles sometimes register better than others, and why this seems to vary if you look at different parts of the registered volume.</div><div class=""><br class=""></div><div class="">-I don’t get any errors during the registration, apart from the occasional “not enough correspondences found”. I normally solve this by slightly relaxing the allowed RANSAC error, and maybe this is part of the problem.</div><div class=""><br class=""></div><div class="">- the literature can be hard indeed. I just found last month’s recording or Stephan’s talk at the 2016 EMBO light sheet course, where he explains some of the parameters in simple terms. Thanks, Stephan!</div><div class=""><a href="https://photolab.mpi-cbg.de/external/LISH2016/23-Preibisch_Stephan.mov" class="">https://photolab.mpi-cbg.de/external/LISH2016/23-Preibisch_Stephan.mov</a></div><div class=""><br class=""></div><div class="">- With a better understanding of the parameters, I’ll continue to try and crack my current data. It is quite clear that I need to optimise the feature based registration, but I’m not there yet. I suspect that perhaps this is also your problem when you say you “struggle with real world samples”?</div><div class=""><br class=""></div><div class="">- I don’t fully get your camera question. What’s the NA of the 20x objective you’re using?</div><div class=""><br class=""></div><div class="">cheers,</div><div class="">Maarten</div><div class=""><br class=""></div><div class=""><br class=""></div><br class=""><div><blockquote type="cite" class=""><div class="">On 2 Sep 2016, at 18:50, Anthony, Neil <<a href="mailto:nantho2@emory.edu" class="">nantho2@emory.edu</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div class="WordSection1" style="page: WordSection1; font-family: Helvetica; font-size: 14px; font-style: normal; font-variant-caps: normal; font-weight: normal; letter-spacing: normal; orphans: auto; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; widows: auto; word-spacing: 0px; -webkit-text-stroke-width: 0px;"><div class=""><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">Hi Maarten, I hope the science treats you well.<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">When I get a registration result that works I feel the overlap in the BDV looks good, although this is a little subjective. Can I ask what kind or errors you get in the log during the registration? I’ve found that I can get it down to less than a pixel and it looks ok.*, **<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""><o:p class=""> </o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">I find that the allowed pixel error helps (I use 1 px), and the significance is less sensitive, but I’ve found 10-15 appears better. Following that the lambda value is better when higher as far as I can tell. I started to look into the literature, but these things are not so accessible on just skimming the papers.<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""><o:p class=""> </o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">I don’t know how much the transformation model and model to regularize with affects things as I haven’t played with those much. Also, I don’t know how the sigma values and thresholds of the bead detection affect the registration, if at all.<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""><o:p class=""> </o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">Thanks<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">Neil<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">* I have a camera that is not properly sampling. I’ve opted to get a larger area and sacrifice resolution on the old ORCA’s we’re using. I have a pixel spacing of 0.64 um, which is approximately the resolution with the 20XW lens. Can I ask if anybody knows how my resolution is affected? I assumed that doubling the field of view in this way would leave me with a resolution ~ 1um…?<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">** My calibration samples, registrations, and<span class="Apple-converted-space"> </span><span class="spelle">MVdecons</span><span class="Apple-converted-space"> </span>are looking pretty nice, but I’m struggling with the real world samples for some reason… I topic for another post once I’ve tried a few things.<o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""> </span><span style="font-family: Calibri, sans-serif;" class=""><o:p class=""></o:p></span></div><div class=""><div style="border-style: solid none none; border-top-color: rgb(225, 225, 225); border-top-width: 1pt; padding: 3pt 0in 0in;" class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><b class=""><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class="">From:</span></b><span style="font-size: 11pt; font-family: Calibri, sans-serif;" class=""><span class="Apple-converted-space"> </span>OpenSPIM [<a href="mailto:openspim-bounces@openspim.org" class="">mailto:openspim-bounces@openspim.org</a>]<span class="Apple-converted-space"> </span><b class="">On Behalf Of<span class="Apple-converted-space"> </span></b>Maarten Hilbrant<br class=""><b class="">Sent:</b><span class="Apple-converted-space"> </span>Monday, August 22, 2016 9:26 AM<br class=""><b class="">To:</b><span class="Apple-converted-space"> </span><a href="mailto:openspim@openspim.org" class="">openspim@openspim.org</a><br class=""><b class="">Subject:</b><span class="Apple-converted-space"> </span>[OpenSPIM] Multiview Reconstruction- optimizing settings?</span><span style="font-family: Calibri, sans-serif;" class=""><o:p class=""></o:p></span></div></div></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">Dear all,<o:p class=""></o:p></span></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">nice to see some questions on the mailing list lately regarding Stephan Preibisch' multview plugin. After attending the EMBO workshop in Dresden in 2014, I've been using the plugin for two years now and it is an essential part of our light sheet workflow. In a nutshell, we image insect embryos, embedded in agarose containing sub-resolution beads, using single-sided illumination with 8 or 16 angles per time point. In order to infer the biological relevance of RNAi phenotypes, we tend to look at many different specimens for few time points, rather than collecting time-lapse data for a small number of specimens. In practice, this means a lot of reconstructing and adjusting settings, although I find that, for most of our datasets, default settings work nicely for bead detection, registration and deconvolution ("Nicely" in this case meaning a significant increase in local contrast). Sometimes, however, registration is sub-optimal (and who knows, maybe it is also for the analyses that I normally quality as "nice''-). In these cases, as a first hint, inspection with the BigDataViewer reveals that the psf's of the different angles don't overlap properly.<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">- Does this sound familiar to other users of the plugin?<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">- What reconstruction parameters do you generally tinker with first?<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">- for example: do you spend a lot of time on bead detection (and subsequent exclusion)?<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">- Do you sometimes reject datasets for multi view reconstruction and if so, on what grounds? For example, I like to image "a lot" of agarose surrounding the embryo in order to have a large set of beads for reconstruction, but am not always very systematic about this. Is this crucial in your data acquisition?<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">any ideas appreciated, basically to make better use of this great tool.<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class=""> <o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">cheers,<o:p class=""></o:p></span></div></div><div class=""><div style="margin: 0in 0in 0.0001pt; font-size: 12pt; font-family: 'Times New Roman', serif;" class=""><span style="font-family: Calibri, sans-serif;" class="">Maarten</span></div></div></div></div></div></div></blockquote></div><br class=""></body></html>