{"id":25029,"date":"2026-11-14T12:08:03","date_gmt":"2026-11-14T12:08:03","guid":{"rendered":"https:\/\/jurn.link\/dazposer\/?p=25029"},"modified":"2026-11-14T14:55:19","modified_gmt":"2026-11-14T14:55:19","slug":"pixel-shift-fix-workflow","status":"publish","type":"post","link":"https:\/\/jurn.link\/dazposer\/index.php\/2026\/11\/14\/pixel-shift-fix-workflow\/","title":{"rendered":"Pixel Shift Fix workflow"},"content":{"rendered":"<p>A medical researcher on Reddit wanted to upscale his old 512px 2000s-era 3D renders to something much better at 2k or 4k, but without changing the facial expression at all. The aim was still a CG 3D look, but better. This prompted me to update my Flux.2 Klein 4B workflow for distribution. It&#8217;s used in ComfyUI Portable and has a Pixel Shift Fix node, which gets the user an exact 1:1 overlay match between the input and the output. 4B has the advantage of a full commercial use permission, meaning his eventual thesis can be redistributed under Creative Commons.<\/p>\n<p><a href=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo.jpg\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo.jpg\" alt=\"\" width=\"640\" height=\"319\" class=\"aligncenter size-large wp-image-25030\" srcset=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo.jpg 1057w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo-300x150.jpg 300w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo-1024x511.jpg 1024w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/man-mouth-medical-demo-768x383.jpg 768w\" sizes=\"auto, (max-width: 640px) 100vw, 640px\" \/><\/a><\/p>\n<p><a href=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute.jpg\"><img loading=\"lazy\" decoding=\"async\" src=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute-1024x466.jpg\" alt=\"\" width=\"640\" height=\"291\" class=\"aligncenter size-large wp-image-25031\" srcset=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute-1024x466.jpg 1024w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute-300x136.jpg 300w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute-768x349.jpg 768w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute-1536x699.jpg 1536w, https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute.jpg 1842w\" sizes=\"auto, (max-width: 640px) 100vw, 640px\" \/><\/a><\/p>\n<p>This ComfyUI Portable workflow should also be useful for those want to quickly emulate a high-quality Octane render, starting from a older or more basic Poser\/DAZ render. The required prompt is in the workflow.<\/p>\n<p>If you only have small 512 or 768px input images to start with, upscale to 2X and sharpen first.<\/p>\n<p>What you do get is a colour shift. In which case, place the input on top of the output in Photoshop, and blend by &#8216;Color&#8217;.<\/p>\n<p><strong>Workflow:<\/strong> <a href=\"https:\/\/jurn.link\/dazposer\/wp-content\/uploads\/2026\/11\/ComfyUI_Flux2_Klein_pixel-fix-distribute.zip\">ComfyUI_Flux2_Klein_pixel-fix-distribute<\/a> <\/p>\n","protected":false},"excerpt":{"rendered":"<p>A medical researcher on Reddit wanted to upscale his old 512px 2000s-era 3D renders to something much better at 2k or 4k, but without changing the facial expression at all. The aim was still a CG 3D look, but better. This prompted me to update my Flux.2 Klein 4B workflow for distribution. It&#8217;s used in [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[9,3,12],"tags":[],"class_list":["post-25029","post","type-post","status-publish","format-standard","hentry","category-freebies","category-poser","category-tutorials"],"_links":{"self":[{"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/posts\/25029","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/comments?post=25029"}],"version-history":[{"count":6,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/posts\/25029\/revisions"}],"predecessor-version":[{"id":25037,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/posts\/25029\/revisions\/25037"}],"wp:attachment":[{"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/media?parent=25029"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/categories?post=25029"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/jurn.link\/dazposer\/index.php\/wp-json\/wp\/v2\/tags?post=25029"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}