Spaces:
Running
Running
Linoy Tsaban
commited on
Commit
•
85762c5
1
Parent(s):
335045f
Update index.html
Browse files- index.html +39 -36
index.html
CHANGED
@@ -317,54 +317,57 @@
|
|
317 |
|
318 |
</div>
|
319 |
</div>
|
320 |
-
|
321 |
-
|
322 |
-
<div class="columns is-centered">
|
323 |
<img
|
324 |
style="max-height:800px; max-width:800px"
|
325 |
src="static/images/semantic_grounding.png"
|
326 |
/>
|
327 |
</div>
|
|
|
|
|
|
|
|
|
|
|
328 |
|
329 |
-
|
330 |
-
|
331 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
332 |
</div>
|
333 |
-
<div class="
|
334 |
-
<
|
335 |
-
<
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
</div>
|
341 |
-
<div class="column">
|
342 |
-
<p>
|
343 |
-
<b>Versatility.</b>
|
344 |
-
LEDITS++ facilitates fine-grained edits and holistic changes such as style transfer.
|
345 |
-
To the best of our knowledge, LEDITS++ is the only diffusion-based image editing method
|
346 |
-
inherently supporting multiple edits in isolation.
|
347 |
-
</p>
|
348 |
-
</div>
|
349 |
-
<div class="column">
|
350 |
-
<p>
|
351 |
-
<b>Precision.</b>
|
352 |
-
LEDITS++’s methodology keeps edits concise and avoids unnecessary deviations
|
353 |
-
from the input image through prefect inversion and use of implict masking
|
354 |
-
(specifically important for editing multiple concepts simultaneously).
|
355 |
-
|
356 |
-
</p>
|
357 |
-
</div>
|
358 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
359 |
|
|
|
|
|
360 |
</div>
|
361 |
-
<video autoplay muted loop playsinline height="100%">
|
362 |
-
<source src="static/videos/objects_styles.mp4"
|
363 |
-
type="video/mp4">
|
364 |
-
</video>
|
365 |
-
</div>
|
366 |
|
367 |
</div>
|
|
|
|
|
|
|
|
|
|
|
368 |
|
369 |
</section>
|
370 |
<section class="section">
|
|
|
317 |
|
318 |
</div>
|
319 |
</div>
|
320 |
+
<div class="columns is-centered">
|
|
|
|
|
321 |
<img
|
322 |
style="max-height:800px; max-width:800px"
|
323 |
src="static/images/semantic_grounding.png"
|
324 |
/>
|
325 |
</div>
|
326 |
+
</div>
|
327 |
+
</div>
|
328 |
+
</section>
|
329 |
+
<section class="section">
|
330 |
+
<div class="container is-max-desktop">
|
331 |
|
332 |
+
|
333 |
+
<div class="columns is-centered has-text-centered">
|
334 |
+
<h2 class="title is-3">Properties of LEDITS++
|
335 |
+
</h2>
|
336 |
+
</div>
|
337 |
+
<div class="columns is-centered">
|
338 |
+
<div class="column">
|
339 |
+
<p>
|
340 |
+
<b>Efficiency.</b>
|
341 |
+
As a parameter-free approach, LEDITS++ does not require any fine-tuning or optimization.
|
342 |
+
In addition we use a recent, fast scheduler altogether making LEDITS++ six times faster than
|
343 |
+
recent DDPM inversion.
|
344 |
+
</p>
|
345 |
</div>
|
346 |
+
<div class="column">
|
347 |
+
<p>
|
348 |
+
<b>Versatility.</b>
|
349 |
+
LEDITS++ facilitates fine-grained edits and holistic changes such as style transfer.
|
350 |
+
To the best of our knowledge, LEDITS++ is the only diffusion-based image editing method
|
351 |
+
inherently supporting multiple edits in isolation.
|
352 |
+
</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
353 |
</div>
|
354 |
+
<div class="column">
|
355 |
+
<p>
|
356 |
+
<b>Precision.</b>
|
357 |
+
LEDITS++’s methodology keeps edits concise and avoids unnecessary deviations
|
358 |
+
from the input image through prefect inversion and use of implict masking
|
359 |
+
(specifically important for editing multiple concepts simultaneously).
|
360 |
|
361 |
+
</p>
|
362 |
+
</div>
|
363 |
</div>
|
|
|
|
|
|
|
|
|
|
|
364 |
|
365 |
</div>
|
366 |
+
<video autoplay muted loop playsinline height="100%">
|
367 |
+
<source src="static/videos/objects_styles.mp4"
|
368 |
+
type="video/mp4">
|
369 |
+
</video>
|
370 |
+
|
371 |
|
372 |
</section>
|
373 |
<section class="section">
|