AI generated images from my custom diffusion models by lavista

View this thread on: hive.blog | peakd.com | ecency.com

hive-158694 · @lavista · Jun 29 '22

$23.12

AI generated images from my custom diffusion models

Hello everyone, hope you are doing fine. I haven’t been able to post much recently as I was training and testing 2 custom AI diffusion models this last week and a half. For more info on how you can create your own AI diffusion model, take a look at this excellent [tutorial](https://peakd.com/hive-158694/@kaliyuga/training-your-own-unconditional-diffusion-model-with-minimal-coding) by @kaliyuga.

#### Here are some images from my first AI diffusion model:


![abs1ups.png](https://files.peakd.com/file/peakd-hive/lavista/AJd95QRSXb9wWZqejMG9M1WU14FR3kaFFCTjaFf8LkbPePbfN4qRtpoMFSMdfPG.png)


![alienups.png](https://files.peakd.com/file/peakd-hive/lavista/AJd9nagw8Sfws7pB7dDd1sviSemocHCeFiHbVP4v8eazjgJEqCmmM3GgWHKwotG.png)


![abs3ups.png](https://files.peakd.com/file/peakd-hive/lavista/AJhTB9wJVWiyLQ35v4x46QCiYr7w457mS2nJ6LW6Tyzamy6WPyYKB2R1ZRo7abH.png)


![ancnec1ups.png](https://files.peakd.com/file/peakd-hive/lavista/AJiuMNr8b8amKTyWz3LLii66nvBvyuE9TBjP5LcgbTZzksFxukH4jdLagmFG4kK.png)


![king3_2ups.png](https://files.peakd.com/file/peakd-hive/lavista/AKRA8xgBEhBKYNA9xRYzvbGuaAMNrWH4RDJqCoGgu6mAgBsUVftmAApjoJe6ivt.png)


![king2ups.png](https://files.peakd.com/file/peakd-hive/lavista/AJebvss3a9ibwCnBaRtLoqsCrkaeXtoprRRe6URZy6D4CG9pnGiUmRc2DSZW4ct.png)

These are all upscaled using Swinir. 
***

#### KIDLIT Diffusion

Here are some images from my 2nd model which we (my wife and I) call kidlit diffusion model. The dataset which I used for training this model was entirely from my wife's digital art which she made using Procreate. Here are some generated images:


![castleups.png](https://files.peakd.com/file/peakd-hive/lavista/ALD6Nps2RMacLZP6pk5Ke168cbTeWXeZEkYL7r86gRqq5roqbD6fn3AE7mTXLy8.png)


![city2ups.png](https://files.peakd.com/file/peakd-hive/lavista/AKDhHWTzG8a8FiVSivSxyYqtyBb83aNQPerPSTMJnj84TPGzEwfVXnRR678qXmP.png)


![city5.png](https://files.peakd.com/file/peakd-hive/lavista/AKVTa7qszpTsTUr8hkEjusWtRAxj3HWRYxN1w8tjMmwpREN2eHoRMuhWicQ3fx8.png)


![dragonups.png](https://files.peakd.com/file/peakd-hive/lavista/AK6W9dv543e1bg2wyuJ9DPfWPHee6i7bkbRcLJ8Hx2A71XaBDSzYcw4NDbKt9Aq.png)


![santaups.png](https://files.peakd.com/file/peakd-hive/lavista/AK56TMarfqvgej8KLHBw4dpthHN7en8bNM12KEWuZZSnSSynWzZPhaijnv8xL6p.png)


![solar2.ups.png](https://files.peakd.com/file/peakd-hive/lavista/AKU4ZtzDZUFgrg5s6a4ZDdQT2dgfScxUfwL7GTmccC5mfYFn1Xet9PBkEFRiL2R.png)


![turtle1.png](https://files.peakd.com/file/peakd-hive/lavista/AK6VmJsog7ZXzJn5H8rdAXVVFDXwjraXWcFHBa3vFJ7cGV3oawodsWsgyFb9B1h.png)


![vas2.png](https://files.peakd.com/file/peakd-hive/lavista/23t7inAwqbbV1B1NshDUizutLfbutEFyoQczcnuXuXk841i6w7PD7v2Hi7kBLpjNfomFf.png)


![vasilisa5.png](https://files.peakd.com/file/peakd-hive/lavista/23u5vyCNzUCVbsGpf27GrC4rS2UwYjncaj2ajUP9VraSGt5ZVv8m6uF6nzR1bdCoMKz3h.png)


![city7.png](https://files.peakd.com/file/peakd-hive/lavista/23tSUEsXfrYjcpJdptYDhh2HyFH7swwRjocDTFduGqgSpAe78Ha5r6qboW414wGb5X2dj.png)


![city8.png](https://files.peakd.com/file/peakd-hive/lavista/AJkQRvTRkkBwt1qTjDGESUXoKUkJ4UcSApBCftEvF7LrM39HBKSgAMsDVXJZH1d.png)

I am absolutely in love with the images produced with the KIDLITDiffusion model!

#### Closing Thoughts
##### When are custom diffusion models needed?
I feel when you have very specific requirements, it is better to train your own model with the appropriate dataset. Kaliyuga's pixel art diffusion notebook is an excellent example. 

On the other hand, if you are just starting out or your needs are more varied, better stick to the open ai diffusion models and explore first. 

##### Some tips while training
I am certainly not an expert but these are things I noticed while training my models.

Train with the highest quality images you have and which you can afford to train on - higher quality images produce larger size training checkpoints and I think they take longer too. The upside is you can probably get away with lesser training steps.

As you can see in the images produced with my first custom model, they are not of great quality, especially after upscaling. This is because I trained the model with just 256x256 sized images. Also, if i try to generate an image with a higher resolution than that, I run into tiling issues.

On the other hand, with my second model, the KIDLIT Diffusion one, I can tell you that most of the images have not been upscaled. I rendered at 1024 x 768 too and there were no tiling issues. 

I think this is because I did not resize images this time while training. Many of the images were 2500x2500 and some even higher. I left them as they were and you can see the quality of the output. What you sow is what you reap is very true here.

Also, it is better to crop your dataset images into square formats.

These are my conclusions from my experiences. I might be very wrong about some things; as I said I am not an expert. If someone notices something wrong here, please let me know in the comments.

That's all for today. Hope you found this useful and entertaining. See you in the next one!
***

**Do you want to generate similar art with the help of AI (or maybe something entirely different)? It is quite simple to get started. You can get started [here](https://peakd.com/hive-158694/@lavista/the-ai-world).**

<img src="https://files.peakd.com/file/peakd-hive/lavista/23tvXtjh2dZZ9oGWtdaccmXtLb6d2FcwXnXkGNm58Xqt71pkmZqcBRfTMNXUW2JfuCWxp.png" alt="divider.png">

<center>
**[linktr.ee](https://linktr.ee/la.vista)**
</center>

<center>
**[My NFTs on NFT Showroom](https://nftshowroom.com/lavista/gallery)**
</center>


<center>
**[My Splinterlands Referral Link](https://splinterlands.com?ref=lavista)**
</center>


![divider.png](https://files.peakd.com/file/peakd-hive/lavista/23tvXtjh2dZZ9oGWtdaccmXtLb6d2FcwXnXkGNm58Xqt71pkmZqcBRfTMNXUW2JfuCWxp.png)

👍 ocdb, curangel, newsflash, usainvote, azircon, ocd, castleberry, galenkp, meritocracy, juliakponsford, bala41288, indiaunited, postpromoter, nateaguila, charlissy, altleft, woelfchen, netaterra, pharesim, litguru, bobinson, roelandp, brotherhood, louis88, cmplxty, bryan-imhoff, thevil, jatinhota, clayboyn, nftshowroom, saboin, thelittlebank, darth-azrael, jlsplatts, dcrops, vachemorte, sunshine, spydo, lemony-cricket, drricksanchez, macchiata, dandesign86, onthewayout, twoitguys, r00sj3, sanderjansenart, eve66, dbddv01, codingdefined, detlev, abh12345, kanibot, hive-168869, nsfw-power, bhdc, jayna, justinw, cooltivar, victartex, actioncats, drwom, egoaquiles, askari, an-man, and 135 others
👎 xcurangel

properties (23)

`author`	lavista
`permlink`	ai-generated-images-from-my-custom-diffusion-models
`category`	hive-158694
`json_metadata`	{"app":"peakd/2022.05.9","format":"markdown","portfolio":true,"tags":["art","surrealism","generativeart","finetuning","customdiffusion","creativecoin","kidlitdiffusion","aiart","ocd","neoxian"],"users":["kaliyuga","kaliyuga.","lavista"],"image":["https://files.peakd.com/file/peakd-hive/lavista/23t7inAwqbbV1B1NshDUizutLfbutEFyoQczcnuXuXk841i6w7PD7v2Hi7kBLpjNfomFf.png","https://files.peakd.com/file/peakd-hive/lavista/AJd95QRSXb9wWZqejMG9M1WU14FR3kaFFCTjaFf8LkbPePbfN4qRtpoMFSMdfPG.png","https://files.peakd.com/file/peakd-hive/lavista/AJd9nagw8Sfws7pB7dDd1sviSemocHCeFiHbVP4v8eazjgJEqCmmM3GgWHKwotG.png","https://files.peakd.com/file/peakd-hive/lavista/AJhTB9wJVWiyLQ35v4x46QCiYr7w457mS2nJ6LW6Tyzamy6WPyYKB2R1ZRo7abH.png","https://files.peakd.com/file/peakd-hive/lavista/AJiuMNr8b8amKTyWz3LLii66nvBvyuE9TBjP5LcgbTZzksFxukH4jdLagmFG4kK.png","https://files.peakd.com/file/peakd-hive/lavista/AKRA8xgBEhBKYNA9xRYzvbGuaAMNrWH4RDJqCoGgu6mAgBsUVftmAApjoJe6ivt.png","https://files.peakd.com/file/peakd-hive/lavista/AJebvss3a9ibwCnBaRtLoqsCrkaeXtoprRRe6URZy6D4CG9pnGiUmRc2DSZW4ct.png","https://files.peakd.com/file/peakd-hive/lavista/ALD6Nps2RMacLZP6pk5Ke168cbTeWXeZEkYL7r86gRqq5roqbD6fn3AE7mTXLy8.png","https://files.peakd.com/file/peakd-hive/lavista/AKDhHWTzG8a8FiVSivSxyYqtyBb83aNQPerPSTMJnj84TPGzEwfVXnRR678qXmP.png","https://files.peakd.com/file/peakd-hive/lavista/AKVTa7qszpTsTUr8hkEjusWtRAxj3HWRYxN1w8tjMmwpREN2eHoRMuhWicQ3fx8.png","https://files.peakd.com/file/peakd-hive/lavista/AK6W9dv543e1bg2wyuJ9DPfWPHee6i7bkbRcLJ8Hx2A71XaBDSzYcw4NDbKt9Aq.png","https://files.peakd.com/file/peakd-hive/lavista/AK56TMarfqvgej8KLHBw4dpthHN7en8bNM12KEWuZZSnSSynWzZPhaijnv8xL6p.png","https://files.peakd.com/file/peakd-hive/lavista/AKU4ZtzDZUFgrg5s6a4ZDdQT2dgfScxUfwL7GTmccC5mfYFn1Xet9PBkEFRiL2R.png","https://files.peakd.com/file/peakd-hive/lavista/AK6VmJsog7ZXzJn5H8rdAXVVFDXwjraXWcFHBa3vFJ7cGV3oawodsWsgyFb9B1h.png","https://files.peakd.com/file/peakd-hive/lavista/23u5vyCNzUCVbsGpf27GrC4rS2UwYjncaj2ajUP9VraSGt5ZVv8m6uF6nzR1bdCoMKz3h.png"]}
`created`	2022-06-29 16:41:42
`last_update`	2022-06-29 16:41:42
`depth`	0
`children`	22
`last_payout`	2022-07-06 16:41:42
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	11.584 HBD
`curator_payout_value`	11.538 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	5,787
`author_reputation`	27,411,778,944,737
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,428,266
`net_rshares`	39,828,566,132,691
`author_curate_reward`	""

vote details (200)

voter	rshares	pct
pharesim	90,475,860,110	7.23%
mammasitta	2,422,845,610	0.7%
roelandp	56,077,886,210	1.75%
onthewayout	12,706,824,441	100%
gikitiki	2,339,335,017	3.61%
sunshine	18,849,259,890	3.5%
bryan-imhoff	36,458,777,582	3.61%
netaterra	91,064,415,023	3.5%
abh12345	10,854,815,487	0.36%
clayboyn	31,387,486,840	25%
keuudeip	545,972,195	6.65%
oleg326756	814,597,670	1.75%
yadamaniart	3,060,637,878	3.5%
justinw	9,255,566,859	2.19%
askari	7,663,156,261	7%
darth-azrael	22,580,237,881	9.9%
darth-cryptic	4,139,998,957	9.9%
detlev	11,014,581,206	1.05%
batman0916	501,440,780	3.32%
rt395	2,121,829,293	2.5%
kaliyuga	637,691,415	100%
newsflash	2,961,393,760,108	5.25%
rawselectmusic	1,362,280,198	3.5%
drwom	8,068,999,413	25%
galenkp	400,617,436,558	10%
dandesign86	14,321,153,900	8%
jayna	9,857,187,459	1.8%
guchtere	642,822,821	3.32%
vachemorte	19,837,220,123	25%
diabolika	1,069,037,104	3.61%
hardikv	659,368,330	5%
juliakponsford	370,962,446,598	50%
auditoryorgasms	700,174,179	25%
diegoameerali	593,980,737	2.16%
scrodinger	1,143,612,551	3.61%
lenasveganliving	2,103,273,209	3.32%
codingdefined	11,140,814,412	3%
ocd	877,125,409,360	7%
macchiata	14,431,841,658	3.5%
twoitguys	12,358,070,552	100%
redrica	2,462,077,018	3.65%
markjason	669,419,463	3.61%
jlsplatts	20,604,713,510	2%
bigdizzle91	818,316,678	3.5%
sayee	5,074,861,012	5%
hanggggbeeee	585,655,553	3.5%
castleberry	425,560,706,327	100%
jatinhota	31,440,035,931	40%
eonwarped	6,494,688,721	0.25%
postpromoter	213,424,050,894	6.65%
bobinson	73,754,375,230	10%
dbddv01	11,273,826,761	100%
paulmoon410	772,146,889	10%
auleo	634,071,706	1.44%
tomatom	689,117,970	3.5%
fourfourfun	720,206,344	2.62%
spydo	17,145,065,397	5%
frames	729,057,460	25%
r00sj3	12,347,821,390	3.5%
sco	1,130,241,759	4.33%
brotherhood	47,528,191,944	3.5%
gabrielatravels	859,137,524	2.8%
bala41288	364,680,185,576	30%
indiaunited	283,148,967,746	10%
edprivat	710,901,489	1.08%
cherryng	867,410,081	1.44%
lemony-cricket	17,087,369,032	3.61%
warpedpoetic	579,400,167	3.32%
ocd-witness	6,165,033,652	7%
bertrayo	2,316,210,344	3.61%
miroslavrc	1,790,668,385	3.5%
azircon	1,110,034,715,643	6.14%
outlinez	1,188,884,707	2.89%
sanderjansenart	11,637,131,848	7%
louis88	40,705,144,456	2.16%
koenau	1,413,256,584	3.61%
greddyforce	673,415,935	2.16%
obsesija	1,404,824,257	3.32%
tijntje	766,724,062	3.61%
manojbhatt	1,073,901,702	10%
saboin	25,024,678,776	2.57%
indigoocean	1,559,088,466	3.5%
veteranforcrypto	925,042,539	2.1%
coloneljethro	2,742,540,338	3.61%
punkblogs	2,779,547,786	30%
nateaguila	175,363,872,678	5%
doctor-cog-diss	3,409,999,030	4.33%
cooltivar	9,018,452,962	5.42%
haccolong	2,260,809,214	3.5%
remotehorst23	2,563,716,957	7.23%
nathyortiz	1,381,955,766	3.5%
cmplxty	40,543,185,410	5.42%
filler	517,385,699	3.61%
ocdb	22,363,698,052,128	6.65%
lifeskills-tv	3,541,687,588	3.32%
litguru	76,024,715,161	100%
hoaithu	802,094,024	2.97%
artmedina	677,842,408	3.5%
voxmortis	1,230,955,856	0.72%
anhvu	571,580,653	2.8%
mobi72	683,603,887	3.5%
nsfw-power	10,597,294,344	7.23%
pradeepdee6	3,328,999,001	2.8%
thelittlebank	23,505,013,169	3.61%
javyeslava.photo	1,205,605,617	3.15%
worksinsane	1,396,644,469	80%
athunderstruck	1,167,307,370	3.5%
thevil	35,448,199,806	3.61%
tsunsica	812,062,910	7%
milky-concrete	5,725,096,804	3.61%
adalger	2,497,063,694	0.72%
devann	5,691,137,259	4%
retrodroid	2,755,039,847	9.9%
variedades	919,934,687	2.66%
robmojo	2,230,414,656	2.16%
khan.dayyanz	6,020,812,647	9%
nftshowroom	26,372,601,943	50%
dawnoner	3,047,954,688	1.08%
olaexcel	972,982,085	5%
photographercr	2,858,003,759	1.33%
squareonefarms	689,286,939	3.5%
epicdice	5,101,696,122	2.16%
yiobri	759,000,759	3.61%
poliwalt10	974,156,918	1.66%
beerlover	1,306,003,849	1.05%
victartex	8,939,209,951	100%
peachymod	1,818,938,356	50%
ssiena	1,425,343,324	5.42%
bala-ag	1,152,716,448	30%
freedomring	677,008,787	1.05%
partitura.ccc	293,244,170	100%
myfreshes	2,786,782,054	3.61%
babytarazkp	950,702,341	10%
kanibot	10,686,807,885	30%
curangel	5,181,453,982,431	7.23%
rootdraws	6,896,553,368	50%
mister.reatard	511,345,588	3.61%
zeesh	1,427,930,281	3.5%
julesquirin	1,658,228,225	10%
dpend.active	498,283,429	1.44%
alienarthive	1,680,876,836	50%
velinov86	2,773,787,419	3.32%
maitt87	1,106,429,918	3.5%
eve66	11,417,108,277	100%
woelfchen	96,411,597,278	100%
actioncats	8,789,314,753	6.3%
gabolegends	467,645,257	1.75%
olaunlimited	5,100,748,229	2.8%
issymarie2	793,870,040	3.5%
vicnzia	639,947,787	4.89%
kawsar8035	1,603,982,120	1.99%
usainvote	2,903,192,975,802	18%
altleft	148,667,379,911	0.36%
hive-168869	10,607,899,689	10%
aemile-kh	7,348,308,543	100%
meritocracy	385,588,617,221	3.61%
stayten	3,142,002,921	5%
godfather.ftw	5,356,423,050	9%
dcrops	19,902,868,531	3.32%
cielitorojo	5,942,611,428	4.65%
marioshelk	963,732,377	100%
an-man	7,658,508,313	20%
limn	3,987,387,373	12.5%
robvector	4,793,511,058	3.61%
drricksanchez	16,167,829,282	3.5%
leveluplifestyle	1,548,350,931	3.5%
jane1289	616,033,048	3.5%
janettbrito76	736,664,957	1.44%
brujita18	831,019,747	3.5%
mattbrown.art	2,084,786,124	25%
lxsxl	2,884,394,798	3.61%
photolovers1	1,833,765,243	5%
partiesjohall	1,111,995,606	7%
nfthypesquad	615,116,814	10%
jessicaossom	5,033,189,744	3.5%
aprasad2325	1,134,634,486	3.32%
abstrads	6,683,019,186	100%
yasern	485,217,573	7%
ivycrafts	841,637,510	3.5%
dlmmqb	6,514,877,764	10%
gehenna08	1,669,032,644	50%
wendy0520	531,000,018	7%
splintercell-01	2,540,427,287	50%
fairyberry	1,090,625,073	3.61%
waivio.curator	1,418,764,566	2.68%
cryptohaytham	2,399,945,983	49%
drexlord	802,849,486	1.75%
tristan.todd	1,205,452,249	7%
yaziris	5,971,025,046	7.23%
mariano123	7,106,786,741	100%
witherbloom	0	100%
akida	2,315,685,952	13%
balaz	5,868,986,108	30%
prosocialise	4,873,202,010	5%
bhdc	10,129,208,217	3.5%
egoaquiles	7,919,934,843	100%
agilecur	6,343,072,419	100%
charlissy	150,643,474,163	100%
xcurangel	-16,376,881	-7.23%
ign1te	2,488,735,495	100%

@castleberry · Jun 30 '22

You and Kali have been doing an excellent job!

👍 lavista, kaliyuga

`author`	castleberry
`permlink`	re-lavista-rea5s0
`category`	hive-158694
`json_metadata`	{"tags":["hive-158694"],"app":"peakd/2022.05.9"}
`created`	2022-06-30 07:50:30
`last_update`	2022-06-30 07:50:30
`depth`	1
`children`	6
`last_payout`	2022-07-07 07:50:30
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	47
`author_reputation`	181,834,289,722,665
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,443,828
`net_rshares`	13,102,451,693
`author_curate_reward`	""

properties (23)vote details (2)

voter	weight	wgt%	rshares	pct	time
kaliyuga	0 B		624,032,033	100%
lavista	0 B		12,478,419,660	100%

@lavista · Jun 30 '22

Thank you:). She is doing much more... She is more like an AI scientist!

👍 kaliyuga

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
kaliyuga	0 B		610,566,305	100%

@castleberry · Jun 30 '22

I am so glad to see these interests taking on a life of their own. I have a feeling we will have AI to help us out creating graphic novels and/or games without having to code as much. The more the models get trained up and as they diversify... it's only a matter of time before they start to converge in cutting corners for bigger projects.

properties (22)

`author`	castleberry
`permlink`	re-lavista-reaa31
`category`	hive-158694
`json_metadata`	{"tags":["hive-158694"],"app":"peakd/2022.05.9"}
`created`	2022-06-30 09:23:27
`last_update`	2022-06-30 09:23:27
`depth`	3
`children`	4
`last_payout`	2022-07-07 09:23:27
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	341
`author_reputation`	181,834,289,722,665
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,445,111
`net_rshares`	0

4 replies

@dbddv01 · Jul 2 '22

Hi @lavista, Thx for your experimental notes about the training. I may have missed the info, but how many pictures did you gather to make model  ? What were your selection criteria for a picture to make part of your dataset (beside HQ size) ? Is colab pro enough ? what was the processing time ? 

@Kaliyuga PAD and now your model are opening new mind bending levels a latent space exploration. This is magic.

👍 lavista, kaliyuga

`author`	dbddv01
`permlink`	reexsq
`category`	hive-158694
`json_metadata`	{"users":["lavista","kaliyuga"],"app":"hiveblog/0.1"}
`created`	2022-07-02 21:46:03
`last_update`	2022-07-02 21:46:03
`depth`	1
`children`	4
`last_payout`	2022-07-09 21:46:03
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	409
`author_reputation`	18,199,928,688,054
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,518,645
`net_rshares`	13,581,375,461
`author_curate_reward`	""

properties (23)vote details (2)

voter	weight	wgt%	rshares	pct	time
kaliyuga	0 B		665,801,228	100%
lavista	0 B		12,915,574,233	100%

@lavista · Jul 3 '22

Hi @dbddv01! I duplicated most of the images by flipping them horizontally, so I had a total of 132 images. 

Regarding the selection criteria, it depends on what you want to do from the model you create. I wanted the children’s illustration / fantasy kind of look. So I choose images accordingly - basically most of my wife's work. Also, I tried to select mostly square format images. I do have some rectangular images in my dataset, but I believe the training program crops the non square images. 

Colab Pro is enough, I did it with Colab Pro. My first model I trained for several days, the training steps count was 152000 when I stopped. With my second model, I started getting decent results at 22000 steps and I stopped training at 32000 steps as I liked the results already. This took only a couple of days of training- I was not training 24hrs/day. Say, I used to train for around 7-8 hours per day for 2.5 days, that's around 24 hours of training for the second model to get good results.

To be honest, with my first model, at around 10000 steps, I wanted to see if it will produce something. I ran it and this was the image produced:

![1.png](https://files.peakd.com/file/peakd-hive/lavista/23swZmfdTUgaUja2Ei3V3pASwosyCssYSMPLZddZgWqCZSqHbJ2RCyqZHFAjHsmX5h5Fx.png)

It indeed felt like magic and I was so satisfied and happy!:).

Looking forward to what you come up with now!

👍 dbddv01

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
dbddv01	0 B		11,134,142,660	100%

@dbddv01 · Jul 3 '22

Thanks a lot for those details. It's looks technically quite affordable, I will have to think of it now. This is exciting !

👍 lavista

`author`	dbddv01
`permlink`	reft9p
`category`	hive-158694
`json_metadata`	{"app":"hiveblog/0.1"}
`created`	2022-07-03 09:05:48
`last_update`	2022-07-03 09:05:48
`depth`	3
`children`	2
`last_payout`	2022-07-10 09:05:48
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	123
`author_reputation`	18,199,928,688,054
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,527,531
`net_rshares`	12,835,693,929
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		12,835,693,929	100%

2 replies

@donnadavisart · Jun 30 '22

Lovely surreal pieces being formulated AI is so interesting and such a great tool for artists. @lavista

👍 lavista

`author`	donnadavisart
`permlink`	reabou
`category`	hive-158694
`json_metadata`	{"users":["lavista"],"app":"hiveblog/0.1"}
`created`	2022-06-30 09:58:06
`last_update`	2022-06-30 09:58:06
`depth`	1
`children`	1
`last_payout`	2022-07-07 09:58:06
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	103
`author_reputation`	280,307,440,291,448
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,445,680
`net_rshares`	12,377,549,003
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		12,377,549,003	100%

@lavista · Jun 30 '22

Hey, thank you so much:)
AI is a great tool for people who are not so artistic too(in the traditional sense), like me:)

properties (22)

@eve66 · Jun 30 '22

Hi, @lavista wow awesome, I need one like this for Horror and collages.

I don't have a current computer, but I hope someday to have one to train my models.

Excellent, congratulations

👍 lavista

`author`	eve66
`permlink`	re-lavista-reb7c3
`category`	hive-158694
`json_metadata`	{"tags":["hive-158694"],"app":"peakd/2022.05.9"}
`created`	2022-06-30 21:21:42
`last_update`	2022-06-30 21:21:42
`depth`	1
`children`	3
`last_payout`	2022-07-07 21:21:42
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	184
`author_reputation`	143,834,346,475,872
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,460,390
`net_rshares`	11,838,923,631
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		11,838,923,631	100%

@lavista · Jul 2 '22

Hi @eve66 . Thank you so much:).

You don’t need a modern computer to do all this. Any device which can run a modern browser and a Google Colab subscription is sufficient. Just follow Kaliyuga's tutorial... The training is done using colab notebooks.

properties (22)

@eve66 · Jul 3 '22

I will try it, Thank you

👍 lavista

`author`	eve66
`permlink`	re-lavista-ref58g
`category`	hive-158694
`json_metadata`	{"tags":["hive-158694"],"app":"peakd/2022.05.9"}
`created`	2022-07-03 00:26:39
`last_update`	2022-07-03 00:26:39
`depth`	3
`children`	1
`last_payout`	2022-07-10 00:26:39
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	24
`author_reputation`	143,834,346,475,872
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,521,180
`net_rshares`	12,657,043,043
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		12,657,043,043	100%

1 reply

@indiaunited · Jun 29 '22

Indiaunited Curation 1656525366388

This post has been manually curated by @bhattg from Indiaunited community. Join us on our [Discord Server](https://discord.gg/bGmS2tE). 

Do you know that you can earn a passive income by delegating to @indiaunited. We share 100 % of the curation rewards with the delegators. 

Here are some handy links for delegations: [100HP](https://hivesigner.com/sign/delegateVestingShares?delegator=&delegatee=indiaunited&vesting_shares=182358.85438660465%20VESTS), [250HP](https://hivesigner.com/sign/delegateVestingShares?delegator=&delegatee=indiaunited&vesting_shares=455897.1359665117%20VESTS), [500HP](https://hivesigner.com/sign/delegateVestingShares?delegator=&delegatee=indiaunited&vesting_shares=911794.2719330234%20VESTS), [1000HP](https://hivesigner.com/sign/delegateVestingShares?delegator=&delegatee=indiaunited&vesting_shares=1823588.5438660467%20VESTS). 

Read our latest [announcement post](https://hive.blog/hive-186042/@indiaunited/indiaunited-2-0-active-again-with-a-lot-more-energy-this-time) to get more information. 

[![image.png](https://files.peakd.com/file/peakd-hive/bala41288/46eaz12N-image.png)](https://discord.gg/bGmS2tE) 

<sub>**Please contribute to the community by upvoting this comment and posts made by @indiaunited.**</sub>

👍 lavista

`author`	indiaunited
`permlink`	indiaunited-1656525366388
`category`	hive-158694
`json_metadata`	{"app":"hiveblog/0.1","tags":["art","surrealism","generativeart","finetuning","customdiffusion","creativecoin","kidlitdiffusion","aiart","ocd","neoxian"]}
`created`	2022-06-29 17:56:06
`last_update`	2022-06-29 17:56:06
`depth`	1
`children`	1
`last_payout`	2022-07-06 17:56:06
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	1,252
`author_reputation`	61,552,599,212,078
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,430,117
`net_rshares`	12,567,995,342
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		12,567,995,342	100%

@lavista · Jun 30 '22

Thank you @bhattg and @indiaunited :)

properties (22)

@kaliyuga · Jun 30 '22

Awesome models!!! I'm so happy to see other people starting to do this :)

👍 lavista

`author`	kaliyuga
`permlink`	re-lavista-reb491
`category`	hive-158694
`json_metadata`	{"tags":["hive-158694"],"app":"peakd/2022.05.9"}
`created`	2022-06-30 20:15:00
`last_update`	2022-06-30 20:15:00
`depth`	1
`children`	1
`last_payout`	2022-07-07 20:15:00
`cashout_time`	1969-12-31 23:59:59
`total_payout_value`	0.000 HBD
`curator_payout_value`	0.000 HBD
`pending_payout_value`	0.000 HBD
`promoted`	0.000 HBD
`body_length`	73
`author_reputation`	243,481,853,306,775
`root_title`	"AI generated images from my custom diffusion models"
`beneficiaries`	`[]`
`max_accepted_payout`	1,000,000.000 HBD
`percent_hbd`	10,000
`post_id`	114,458,484
`net_rshares`	11,608,046,866
`author_curate_reward`	""

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
lavista	0 B		11,608,046,866	100%

@lavista · Jul 2 '22

Thank you so much:). Like I have said before, you were an inspiration!

👍 kaliyuga

properties (23)vote details (1)

voter	weight	wgt%	rshares	pct	time
kaliyuga	0 B		651,505,086	100%